Brain Network Dysfunction in Post-Stroke Delirium and Spatial Neglect: An fMRI study.

Abstract

Background and Purpose:

Delirium, an acute reduction in cognitive functioning, hinders stroke recovery and contributes to cognitive decline. Right-hemisphere stroke is linked with higher delirium incidence, likely, due to the prevalence of spatial neglect (SN), a right-brain disorder of spatial processing. This study tested if symptoms of delirium and SN after right-hemisphere stroke are associated with abnormal function of the right-dominant neural networks specialized for maintaining attention, orientation, and arousal.

Methods:

Twenty-nine participants with right-hemisphere ischemic stroke undergoing acute rehabilitation completed delirium and SN assessments and functional neuroimaging scans. Whole brain functional connectivity of 4 right-hemisphere seed regions in the cortical-subcortical arousal and attention networks was assessed for its relationship to validated SN and delirium severity measures.

Results:

Of 29 patients, 6 (21%) met the diagnostic criteria for delirium and 16 (55%) for SN. Decreased connectivity of the right basal forebrain to brainstem and basal ganglia predicted more severe SN. Increased connectivity of the arousal and attention network regions with the parietal, frontal, and temporal structures in the unaffected hemisphere was also found in more severe delirium and SN.

Conclusions:

Delirium and SN are associated with decreased arousal network activity and an imbalance of cortico-subcortical hemispheric connectivity. Better understanding of neural correlates of post-stroke delirium and SN will lead to improved neuroscience-based treatment development for these disorders.

Clinical Trial Registration Information:

http://www.clinicaltrials.gov; NCT03349411

Introduction

Delirium, previously known as acute confusional state, is a dangerous complication after stroke and efforts to improve understanding of its brain mechanisms have both preventive and therapeutic value. Delirium, defined as a sudden decline or fluctuation in attention, orientation, arousal, and memory, occurs in 23% (95% CI, 17%-28%) of stroke survivors.¹ It increases the length of hospital stays and mortality^2-4, and can undercut neuro-recovery rates among stroke inpatients, leading to prolonged rehabilitation, increased risk of in-hospital falls, and reduced cost-effectiveness of rehabilitation medicine.⁵ A substantial percentage of delirium cases may be preventable.⁶ However, neural mechanisms are not yet being manipulated to prevent, manage, or treat delirium, because we lack detailed information about the brain correlates of this disorder. Only a few functional^7-9 and structural^10-14 neuroimaging studies have been carried out to investigate the brain structures affected in delirium, and little is known about the brain networks impaired in delirium after stroke.

Right-brain stroke is reported to increase delirium incidence^2,15,16 and this relationship may be mediated by another disorder of attention and spatial processing, called spatial neglect (SN), shown to be an independent delirium predictor.^17-19 SN is defined as asymmetric attention and action after a brain lesion, which causes functional disability.^20,21 Considering that attention deficits represent a prototypical feature of both delirium and SN, we previously proposed that the increased risk for delirium in stroke patients with SN arises because of the abnormal function of the right-dominant neural networks specialized for maintaining attention and orientation.²²

In addition, the dysfunction of the central nervous system’s arousal network likely plays an important role in the onset and development of delirium. The ascending reticular activating system (ARAS), initiates and maintains wakefulness and arousal. Its relationship to the cardinal symptoms of delirium: disturbance of arousal and attention, and dysregulation of sleep-wake cycle, has been discussed decades ago by Ross²³ and Mesulam and colleagues.²⁴ The ARAS consists of several neuronal circuits which carry neurotransmitter-specific projections from brainstem source nuclei to the thalamus, hypothalamus, basal forebrain, and the cerebral cortex.²⁵ Thus, reduced availability of some neurotransmitters (e.g., acetylcholine) and increased release of others (e.g., dopamine) in delirium²⁶ may be associated with a dysfunction of the ARAS. Direct evidence for the role of the arousal network in delirium comes from a functional neuroimaging study monitoring patients during and after an episode of delirium.⁷ Delirious patients, compared to healthy participants, had a disruption of connectivity between components of the ARAS (basal forebrain and the thalamus). Structural integrity of these components measured with diffusion tensor imaging could also be used to reliably predict delirium after elective surgery.¹⁴ Little is known about the functioning of the ARAS and of the attention and orientation networks in patients with stroke who are at risk for SN and delirium. We addressed this knowledge gap in this resting-state functional MRI (fMRI) study.

Resting state fMRI (rs-fMRI) is a promising tool for studying functional connectivity associated with acute brain failure in delirium, but it presents unique challenges. Patients with stroke, SN, and/or delirium may feel drowsy or move excessively during a neuroimaging scan. An additional challenge is the variability of SN and delirium over short time periods. To address these challenges, we employed robust motion correction of neuroimaging data based on independent component analysis.²⁷ While we did not explicitly monitor drowsiness, which should be addressed in future work, we alerted patients at the start of each rs-fMRI run. We also examined SN and delirium on a continuum of severity from little to severe impairments using high-reliability and validity functionally-relevant assessments.^28,29 Both delirium and SN severity were examined in the same mixed linear-effects model testing the ability of these variables to predict patterns of spontaneous brain activity in the attention/orientation and arousal networks. We hypothesized that delirium and SN are associated with a persisting disruption of functional connectivity among brain areas within neural networks specialized for maintaining attention, orientation, and arousal. We tested this hypothesis by examining whole-brain functional connectivity of two seed regions within the ARAS [thalamus (Thal) and basal forebrain (BF)] and two seeds within the attention and orientation networks [superior (SPL) and inferior parietal lobules (IPL)].

Methods

Participants

The data that support the findings of this study are available from the corresponding author upon reasonable request. The data were collected between August 2017 and December 2020 as part of a prospective observational cross-sectional neuroimaging investigation of SN and delirium in stroke participants undergoing inpatient rehabilitation. The sample size (N=30) required to achieve 80% power was estimated a priori using a pilot fMRI study. To achieve the required level of recruitment 33 eligible participants were enrolled. Four participants did not complete study assessments due to being discharged or transferred to an acute care facility and unable to continue the study. The final sample included 29 right-hemisphere stroke survivors (20 women and 9 men; Table 1). Participants ages 18-100 were included in the study if they had a first-ever ischemic stroke; no clinically-diagnosed left stroke and no prior neurological disorders, were within the first 3 months post-stroke, spoke English, and had no MRI contraindication. All participants provided a written informed consent in the presence of a witness according to the local IRB guidelines (See Supplemental Materials for the informed consent procedure). This study follows STROBE³⁰ reporting guidelines.

Table 1.

Participant Characteristics

Measure (Units)	N=29
	Mean	Median	SD	Range	IQR*
Age (in years)	65.31	67	10.65	41-86	11
Years of education (in years)	13.57	12	2.69	9-19	3
Lesion size (in cm3)	41.25	26.06	47.22	0.79-154.13	58.0
Days post stroke	24.59	16	13.78	10-66	14
Number of admission medications (count)	12.74	13	5.63	5-31	8
Patients on sedatives (n, %)	4 (14%)
Patients on other psychoactive drugs (e.g., SSRIs, anticonvulsants) (n, %)	7 (24%)
Visual field defects, confrontation testing (n, %)	6 (21%)
CBS† via the KF-NAP‡ (score, 0-30)	6.51	3.78	7.35	0-22.22	9.42
BIT-c§ (score, 0-146)	110.52	129	38.82	24-146	55
SN, BIT-c < 130 (n, %)	16 (55%)
CAM-S∣ (average score, 0-19)	3.74	3	2.40	0-9.67	2.76
Delirium, 3 diagnostic features on any CAM (n, %)	6 (21%)
Subsyndromal delirium, 2 diagnostic features on any CAM (n, %)	13 (45%)
Boston Naming Test (% correct)	72.84	80	21.48	20-100	20
Geriatric Depression Scale (score, max score = 30)	8.32	9	6.99	0-22	11.3
Hopkins Verbal Learning Test
Immediate recall (36)	17	19	5.81	5-24	8.50
Delayed recall (12)	4.52	5	3.42	0-12	5
Test recognition index: True – false positives (12)	9.22	10	3.24	0-12	3
Motor Function (strength score)
Proximal left upper extremity (5)	2.61	3	2.04	0-5	4
Distal left upper extremity (5)	2.54	3.5	2.03	0-5	4
Proximal left lower extremity (5)	2.93	4	1.86	0-5	3
Distal left lower extremity (5)	2.93	4	1.98	0-5	3.5

^*Interquartile Range

^†Catherine Bergego Scale

^‡Kessler Foundation Neglect Assessment Process

^§Behavioral Inattention Test-conventional

^∣Confusion Assessment Method-Severity

Materials

Delirium Assessment.

Confusion Assessment Method (CAM)³¹ and CAM-S²⁸ were used for delirium assessment. The CAM is a validated tool assessing 4 diagnostic and 6 associated delirium symptoms and CAM-S is used to score severity of the symptoms assessed on the CAM (range 0-19, higher scores represent worse impairment). Patients with >2 diagnostic symptoms, including acute onset and inattention, are considered delirium positive. Patients with 2 diagnostic symptoms are considered subsyndromal, with intermediate prognoses compared to non-delirious and delirious patients.³²

SN Assessment.

SN assessment was performed using the validated Behavioral Inattention Test – conventional (BIT-c).³³ A score of <130 on the BIT-c represents clinical cutoff for SN. Limitations of daily life function due to SN were assessed using the Catherine Bergego Scale (CBS), administered via the Kessler Foundation-Neglect Assessment Process (KF-NAP)³⁴. A confrontation test was used to assess visual field defects.

Mental Status Assessments.

Patients were screened with a bedside neuropsychological scoring procedure using the Florida Mental Status Exam (FMSE)³⁵. Because depression was associated with delirium^6,36, patients were evaluated for depression with the Geriatric Depression Scale (GDS).³⁷ We obtained information about the patients’ medications, as some drugs may precipitate delirium.⁶ The number of medications and the frequency of use of sedative or otherwise psychoactive drugs is noted in Table 1. Additional details about assessments appear in Supplemental Material.

Procedure

Behavioral Assessment.

Behavioral assessments were administered by trained research staff within 1 week of study participation. Delirium was assessed 1-3 times in each participant and averaged severity scores were used in the analysis. CAM testing alternated between left and right side, whenever the position directly in front of the patient was not available (e.g., patient in bed). SN was assessed once with materials positioned at the patient midline. BIT-c scores were converted to visuospatial impairment scores by subtracting them from 146. Both CAM-S and BIT-c scores were converted to z-scores (mean centered and scaled) for the neuroimaging analysis. CBS via KF-NAP was assessed by occupational therapists (unaware of BIT-c score) within 5 business days of admission as part of clinical care (See Fig.1 for timeline).

Figure 1.

Study timeline. Patients with first-ever ischemic stroke were enrolled 24.6 days post-stroke. Functional disabilities due to spatial neglect were assessed using KF-NAP within 5 business days of hospital admission. Delirium and spatial neglect assessments were conducted within 1 week and MRI within 10.4 days of study enrollment.

MRI Acquisition.

All participants underwent MRI on a 3T Siemens Skyra Scanner using a 20-channel head/neck coil. The imaging occurred 10.43 days (SD=11.06) after consent in the same week or immediately following behavioral assessment. The following images were acquired: high-resolution T1-weighted Magnetization-Prepared Rapid Gradient-Echo (MPRAGE), FOV=256mm, TR=2100ms, TE=3.43ms, flip angle=9°, n slices=256, 1mm³ voxels; T2-weighted Fluid Attenuated Inversion Recovery (FLAIR), FOV=256mm, TR=9000ms, TE=91ms, flip angle=150°, n slices=50, 1x1x3mm voxels; and T2*-weighted gradient-echo echo-planar imaging (EPI), FOV=220 mm, TR=2000ms, TE=30ms, flip angle=70°, n slices=32, 2.3x2.3x3mm voxels, 0.6mm gap, 240 volumes. The T2*-weighted EPI images were acquired at rest with eyes closed. Participants were instructed to stay awake at the start of this scan and, depending on self-reported ability to tolerate the scan, completed 1 (N=9) or 2 runs (N=20).

Lesions.

Lesions were mapped using manual segmentation and automated intensity-based voxel selection in FSLeyes.³⁸ Approximate lesion location was identified from medical history. T1-weighted and T2-FLAIR images were used for identification of voxels with abnormal intensity. To avoid warping of the lesion area during registration to the anatomical template, we applied cost-function masking of the input image using the lesion mask. Lesioned areas on the T2*-weighted EPI were excluded from the analysis, so that none of the functional connectivity measures are based on structurally damaged regions. Log-transformed lesion volume was used as a control variable to account for injury severity.

Analysis

Behavioral Data.

Rates of delirium and SN were noted. To study the relationship between delirium and SN symptoms a partial correlation analysis was conducted between BIT, KF-NAP and CAM-S scores, controlling for age, days post-stroke (to consent), GDS, and lesion volume.

Neuroimaging Data.

T2*-weighted resting-state scans were brain extracted using BET³⁹, motion corrected using MCFLIRT, an automated tool for intra-modal motion correction⁴⁰, and smoothed with 5mm FWHM kernel. Affine and non-linear registration matrices were computed for registration to MNI152 2mm brain template using FLIRT and FNIRT with cost-function masking and intermediate registration to the participant’s T1-weighted anatomical image.^38,40. Further removal of motion components was achieved using ICA-based strategy for Automatic Removal of Motion Artifacts [ICA-AROMA].²⁷ ICA-AROMA identifies a set of theoretically motivated temporal and spatial independent components associated with head motion and removes them by regression. Additionally, we performed nuisance regression of CSF and WM signals (excluding the brainstem) and Gaussian high-pass temporal filtering (Gaussian-weighted least-squares straight line fitting), with 100 s FWHM.

Eigen-timeseries were extracted from 4 anatomically-defined seed regions of interest (ROI) within the functional neural networks for arousal (right Thal and BF)^25,41 and attention (right IPL and SPL).^42,43 The ROIs were defined as spheres (r=10mm) around coordinates in MNI space (rIPL: x=50, y=−53, z=25; rSPL: x=31, y=−47, z=61; rThal: x=10, y=−16, z=8; rBF: x=12, y=−2, z=−8). To ensure that choice of ROI radius could not account for the present findings, an analysis with smaller ROI radius (5mm) was also conducted, showing convergent findings (see Suppl. Fig.5). For each participant, the ROI timeseries were regressed on pre-processed whole brain activity using a model-based fMRI analysis tool FEAT.⁴⁴ This resulted in 4 spatial maps per participant per run showing coordinated patterns of brain activity between the 4 seed ROIs and the rest of the brain. These maps were aggregated across runs and participants in order to derive group-level functional connectivity patterns for each seed ROI. Fixed-effects modeling was used to average between runs of a single participant and mixed-effects modeling (with participants as a random-effects factor) was used for group analysis. To identify brain connectivity patterns that progressively associate with increasing severity of delirium and SN, we chose to treat these disorders on a spectrum. Therefore, at the group level, we computed both the average whole-brain functional connectivity for each seed ROI and functional connectivity associated with 2 regressors representing symptom severity (centered and scaled BIT-c and CAM-S). This was done within a single model in order to account for independent contributions of these regressors to functional connectivity variance. All z (Gaussianised t/F) statistic images were thresholded non-parametrically using clusters determined by Z>2.3 and a (corrected) cluster significance threshold of p=0.05.

Results

Behavioral Data

Delirium was present in 6 (21%) and subsyndromal delirium in 13 patients (45%), based on at least one of the 1-3 available CAM assessments. SN was present in 16 patients (55%) (Table 1). When controlling for age, days post stroke, GDS, and lesion volume, CAM-S scores were significantly associated with KF-NAP scores (r=.55, p<.02; Fig2.B), but not with BIT-c scores (r=−.26, ns). CAM-S did not significantly correlate with the total number of medications or the GDS score. As expected, BIT-c and KF-NAP scores (measures of SN) were significantly correlated (r=.51, p<.05).

Figure 2.

Results overview. A. Stroke lesion overlay. Areas of greater lesion overlap appear in green-red. B. Group rsFC associated with delirium and SN severity. C. Sample correlation plots of significant rsFC-syndrome severity associations. Plots constructed using peak voxel beta-weights from participant-level analysis and BIT-c and CAM-S z-scores, shaded areas represent 95% confidence intervals. Significance was established using group-level general linear model with non-parametric cluster-level correction (p<0.05), simple correlation and p-values are also shown. AG – angular gyrus, Am – amygdala, aSTG – anterior superior temporal gyrus, BF – basal forebrain, Calc – calcarine cortex, CC – cingulate cortex, Ins – insula, IPL – inferior parietal lobule, LG – Lingual gyrus, LO – lateral occipital cortex, MFG – middle frontal gyrus, pOP/cOP – parietal/central operculum, ParaCing – paracingulate cortex, PreCu – precuneus, SMG – supramarginal gyrus, SPL – superior parietal lobule, Thal – thalamus.

Neuroimaging Data

Basal Forebrain.

Group analysis revealed significant resting-state functional connectivity (rsFC) of the rBF with bilateral basal ganglia (caudate nucleus, putamen, and pallidum), thalamus, parahippocampal and cingulate gyri, cerebellum, mesencephalic reticular formation, right hippocampus and left insula, middle frontal, and angular gyri (Suppl. Fig.1, Suppl. Table 1). RsFC of rBF with left precuneus and angular gyrus/lateral occipital cortex was positively associated with severity of delirium symptoms (Fig.2A,C, Suppl. Fig.1B, and Suppl. Table 1). RsFC of rBF with left calcarine and supracalcarine (visual) cortex was positively associated, whereas rBF connectivity with left brainstem, putamen, pallidum, amygdala, insula, and parietal operculum was negatively associated with SN severity (Fig. 2A,C).

Thalamus.

RsFC of the rThal included bilateral brainstem, cerebellum, paracingulate, cingulate, and parahippocampal gyri, frontal pole, left hippocampus, angular gyrus/lateral occipital cortex, insula, middle frontal/precentral, posterior inferior temporal, supramarginal, and lingual gyri (Suppl. Fig.2A, Suppl. Table 2). RsFC of rThal with left angular gyrus/lateral occipital cortex and anterior superior temporal gyrus was associated with greater SN severity (Fig.2A,C, Suppl. Fig.2B, Suppl. Table 2).

Inferior Parietal Lobule.

RsFC of rIPL included large clusters in the precuneus, left posterior parietal, paracingulate and medial frontal cortex, and smaller clusters in the right posterior parietal and temporal cortex. Significant rsFC was also found with the cerebellum, brainstem, and left thalamus (Suppl. Fig.3A, Supp. Table 3). Connectivity of the rIPL with left central opercular cortex (inferior frontal, precentral and postcentral gyri), insula and supramarginal gyrus was positively correlated with delirium symptom severity (Fig.2A, Suppl. Fig.3B, Supl. Table 3). Connectivity of the rIPL with bilateral precentral, postcentral, cingulate and left middle frontal gyri was positively associated with SN severity. Connectivity of the rIPL with bilateral precuneus, left lingual and angular gyri/lateral occipital cortex was negatively associated with SN severity (Fig.2A, Suppl. Fig.3C, Suppl. Table 3).

Superior Parietal Lobule.

RsFC of rSPL included bilateral superior parietal, lateral occipital and mid/anterior cingulate cortex, lingual gyrus, cerebellum, right lateral occipital/inferior temporal cortex, thalamus, frontal pole/middle frontal and left postcentral gyri (Suppl. Fig.4A, Suppl. Table 4). Connectivity of rSPL with bilateral precuneus, paracingulate and parahippocampal gyri, cerebellum, and ventral striatum was positively associated with SN severity (Fig.2A, Suppl. Fig.4B, Suppl. Table 4).

Discussion

The goal of this study was to investigate the relationship between SN and delirium symptoms and patterns of rsFC in right-hemisphere stroke. Motivated by evidence of symptom overlap between SN and delirium²², including disturbance of attention and lack of vigilance, we examined the functioning of two brain networks with overlapping specialization for maintaining arousal, orientation, and attention. Our hypothesis was that disruption of these networks underlies the onset of delirium and SN. We provide support for this hypothesis and implicate increased connectivity of these networks with frontal, parietal, and temporal structures of the unaffected hemisphere in more severe delirium and SN.

The first examined network was the ARAS, which supports wakefulness and arousal.^21,45,46 Consistent with prior studies, ARAS connectivity included the brainstem, bilateral thalamus, basal forebrain, hypothalamus, basal ganglia, and cingulate gyrus (Suppl. Fig.1A&2A). Increased ARAS connectivity with the left calcarine, lateral occipital, angular, and precuneal cortex was associated with greater delirium and SN severity. This is in line with prior animal models of attention and oculomotor behavior. In cats, unilateral parieto-occipital lesions result in hyper-activation of the unaffected hemisphere and profound deficits in contralesional orienting.^47,48 This effect is mediated by cortical-subcortical connectivity. Orienting deficits resolve after a second lesion to the unaffected hemisphere as shown in animal^47,48 and human case studies⁴⁸. Thus, increased ARAS connectivity with occipito-parietal areas in the left unaffected hemisphere appears maladaptive. In contrast, increased connectivity within the ARAS (rBF to brainstem, basal ganglia, and amygdala) was linked with reduced SN severity. This is consistent with prior studies showing that increased alertness improves post-stroke SN.⁴⁹ We also found an inverse relationship between rBF-left insula and rBF-parietal operculum connectivity and SN severity, which warrants further investigation.

We next examined the fronto-parietal brain networks that modulate attention. The observed rIPL connectivity matched with the map for ventral attention network in healthy young adults (Fig.8, Hacker et al.⁵⁰), except that in our participants, the left, compared to the right, side of the brain was over-represented. RSPL connectivity matched with the connectivity expected for dorsal attention network (Fig.8, Hacker et al.⁵⁰), but with greater left-hemisphere involvement. The overreliance on the left non-lesioned hemisphere is not surprising, if unilateral right-brain stroke resulted in reduced availability of healthy brain tissue in the right hemisphere.

The functional connectivity-behavior associations showed increased connectivity of the attention network seeds with the left hemisphere and with several bilateral cortical and subcortical areas in more severe SN and delirium. We found that higher rIPL connectivity with left parietal, temporal, and frontal regions correlated with increased delirium and SN severity. Increased connectivity with bilateral midline structures (e.g., dorsal precentral and postcentral gyri) was associated with increased SN severity. Furthermore, increased rSPL connectivity with left anterior temporal cortex, bilateral posterior parietal areas near brain midline, basal ganglia, parahippocampal gyrus and cerebellum correlated with worse SN. Attentional/oculomotor orienting is subserved by a network of cortico-subcortical connections, where mutual inhibition provides a means to selectively distribute attention within different spatial reference frames.^47,48,51 Our results show that more severe SN is coupled with inefficient connectivity patterns. For example, increased rSPL connectivity with bilateral basal ganglia would cue bilateral oculomotor nuclei and provide conflicting signals about orienting to the left and the right side of space simultaneously. In contrast, less severe SN was associated with higher rIPL connectivity to bilateral visual/visual association areas, which may indicate occipito-parietal white matter preservation. We previously showed that the integrity of this white matter was a predictor of SN-associated reading errors.⁵²

This study suggests that connectivity of regions within ARAS and attention and orientation networks is associated with delirium and SN severity. We found support for our hypothesis of disrupted connectivity affecting these networks by showing an inverse relationship between ARAS connectivity and SN severity and showing that ARAS-mediated connectivity with left-hemisphere occipito-parietal areas was implicated in more severe SN and delirium. This is consistent with animal studies demonstrating that lesions of the ARAS and its frontal projection sites can cause SN and hypokinesia in monkeys.^53,54 Similarly, ARAS was shown to be dysfunctional in delirium. EEG patterns associated with decreased activity in the ARAS⁵⁵ were present in greater proportion among delirious patients.^56,57 We also showed that dorsal and ventral attention network connectivity was shifted to the left hemisphere and this was associated with more severe delirium and SN. This is consistent with studies showing that dysfunction of right dorsal and ventral parietal cortices are common in stroke patients with SN^58-60 and in some patients with post-stroke delirium.¹³ Furthermore, our findings provide a more nuanced understanding of the previously established relationship between delirium and SN. Delirium symptom severity was more associated with severity of functional limitations due to SN (measured with KF-NAP), than with deficits in visuo-spatial processing on paper and pencil tasks (BIT).

Study Limitations and Future Directions

The choice of methodology should be considered when interpreting these results. Given individual variability among stroke survivors, the modest sample size and different number of rs-fMRI runs is a limitation. We did not include a non-stroke control group or compare patient groups with and without delirium and SN. We chose to treat delirium and SN symptoms on a spectrum to identify brain connectivity patterns that progressively associate with increasing severity of these disorders. Future large-scale studies should examine connectivity patterns and their behavioral correlates in large groups of patients with and without delirium and SN. While participants were alerted at the start of each resting-state run, we did not collect simultaneous data on sleep states (e.g., electroencephalogram), which should be addressed in future work.

The prevalence of delirium is greater in the acute post-stroke period.^5,6 It is possible that some or all delirium symptoms present acutely have resolved by the time our participants were admitted to acute rehabilitation and took part in the study. However, it is now increasingly recognized that persistent delirium is common and may lead to irreversible cognitive decline.^61,62 Delirium among patients during inpatient rehabilitation is one of the most common reasons for transfers to acute care hospital⁵. The occurrence and persistence of delirium in this population is understudied. Similarly, the incidence of SN and the associated functional disability is alarmingly high even upon discharge from acute rehabilitation.³⁴ Our findings further suggested that functional disability due to spatial neglect is associated with presence of delirium symptoms. Thus, more studies are needed on the incidence of these disorders and their association with rehabilitation outcomes.

Non-Standard Abbreviations and Acronyms

ARAS

Ascending Reticular Activating System

rBF

Right Basal Forebrain

fMRI

Functional Magnetic Resonance Imaging

rIPL

Right Inferior Parietal Lobule

KF-NAP

Kessler Foundation Neglect Assessment Process

SN

Spatial Neglect

rsFC

Resting State Functional Connectivity

rSPL

Right Superior Parietal Lobule

rThal

Right Thalamus