Severe Cognitive Impairment in Aneurysmal Subarachnoid Hemorrhage: Predictors and Relationship to Functional Outcome

Abstract

Background

Cognitive impairment is common after aneurysmal subarachnoid hemorrhage (SAH). However, compared to predictors of functional outcome, meaningful predictors of cognitive impairment are lacking.

Objective

Our goal was to assess which factors during hospitalization can predict severe cognitive impairment in SAH patients, especially those who might otherwise be expected to have good functional outcomes. We hypothesized that the degree of early brain injury (EBI), vasospasm, and delayed neurological deterioration (DND) would predict worse cognitive outcomes.

Methods

We retrospectively reviewed SAH patient records from 2013–2019 to collect baseline information, clinical markers of EBI (Fisher, Hunt-Hess, and Glasgow Coma scores), vasospasm, and DND. Cognitive outcome was assessed by Montreal Cognitive Assessment (MoCA) and functional outcomes by modified Rankin Scale (mRS) at hospital discharge. SAH patients were compared to non-neurologic hospitalized controls. Among SAH patients, logistic regression analysis was used to identify predictors of severe cognitive impairment defined as a MoCA score <22.

Results

We screened 288 SAH and 80 control patients. Cognitive outcomes assessed via MoCA at discharge were available in 105 SAH patients. Most of these patients had good functional outcome at discharge with a mean mRS of 1.8±1.3. Approximately 56.2% of SAH patients had MoCA scores <22 compared to 28.7% of controls. Among SAH patients, modified Fisher scale was an independent predictor of cognitive impairment after adjustment for baseline differences (OR 1.638, p=0.043). MoCA score correlated inversely with mRS (r=−0.3299, p=0.0006); however, among those with good functional outcome (mRS 0–2), 48.7% still exhibited cognitive impairment.

Conclusions

Severe cognitive impairment is highly prevalent after SAH, even among patients with good functional outcome. Higher modified Fisher scale on admission is an independent risk factor for severe cognitive impairment. Cognitive screening is warranted in all SAH patients, regardless of functional outcome.

INTRODUCTION

Aneurysmal subarachnoid hemorrhage (SAH) is a devastating form of hemorrhagic stroke resulting from the rupture of intracranial aneurysms. SAH accounts for 5% of strokes annually, but up to 27% of all stroke-related years of potential life lost due to long-term disability and mortality.^1–3 SAH additionally affects younger patient populations compared to ischemic stroke, and thus has a significant impact on ability to return to work and years of productive life. One significant cause of long-term disability is cognitive impairment, estimated to occur in at least 50% of all SAH patients.^4–5 Cognitive impairment mainly affects domains of memory, executive function, and language.^6–7 Most outcome scales used in SAH, including the modified Rankin scale (mRS) and Glasgow Outcome Scale (GOS), focus mainly on motor and other neurological deficits. However, they are not particularly sensitive in detecting cognitive decline. In addition, risk factors and mechanisms leading to cognitive impairment in SAH patients are relatively unknown, making it difficult to predict which patients may be at higher risk and benefit from neurocognitive interventions.

Neurological damage after SAH can be divided into two phases – early brain injury (EBI), which occurs in the first 72 hours, and delayed neurological deterioration (DND) which can occur anywhere from 3–14 days after aneurysm rupture. Various mechanisms likely influence the severity of EBI, including neuroinflammation, microvascular dysfunction, microthrombosis, and electrophysiologic abnormalities including cortical spreading depolarization and epileptiform discharges.^8–9 Clinically, EBI is associated with several factors including location and severity of bleed, cerebral edema, hydrocephalus, presence of focal neurological deficits or infarction, and level of consciousness.¹⁰ From the practical stand point, the Hunt-Hess, modified Fisher, and Glasgow Coma scales (GCS) have been used as surrogate markers of SAH severity.^10–12

Most studies in SAH have focused on the role of delayed cerebral vasospasm and its contribution to neurologic disability. Cerebral vasospasm occurs in 70% of patients.¹³ However, results of several studies have suggested that vasospasm is not the sole factor driving chronic impairments.^8,9,14,15 More recently, DND has emerged as a major driver of long-term functional outcome.¹⁶ DND is a phenomenon related to but distinct from cerebral vasospasm that occurs in up to 30% of patients and presents with focal neurological deficits that may progress to infarction.³ Thus, DND is defined by a late neurologic deterioration after SAH but does not require the presence of vasospasm. However, despite known associations with functional outcome, the relationship between cerebral vasospasm and DND in driving poor cognitive outcome is less well-known.

In this study, we investigated the contribution of SAH-associated EBI and DND to severe neurocognitive impairment using the Montreal Cognitive Assessment (MoCA). The MoCA is a 30-point cognitive screening tool used to assess cognitive impairment across a range of neuropsychiatric conditions. The MoCA covers several cognitive domains including visuospatial memory, executive function, memory, attention, language, abstract reasoning, delayed recall, and orientation to time and place.^7,17,18 We hypothesized that SAH survivors, even those with low-grade SAH, would show increased rates of severe cognitive impairment compared to controls, and that those with a higher degree of EBI, cerebral vasospasm, and DND would exhibit more cognitive impairment.

METHODS

Population

This study was approved by the institutional review board of the University of Illinois Hospital & Health Sciences System. The study population included all patients greater than 18 years of age with a diagnosis of aneurysmal SAH admitted to our hospital between January 2013-July 2019. Cases were retrospectively identified by ICD-9 (430) and ICD-10 (I6000, I6001, I6002, I6010, I6011, I6012, I602, I6030, I6031, I6032, I604, I6050, I6051, I6052, I606) discharge codes. SAH was confirmed by CT head or lumbar puncture. Cerebral aneurysms were confirmed by digital subtraction angiography and/or computerized tomography angiography. Patients with non-aneurysmal SAH and unable to complete the MoCA were excluded from our study. Reasons for not being able to complete the MoCA included patient refusal, death, comfort measures only, or medical conditions that would prevent accurate assessment of cognitive status such as intubation, comatose state, agitation, communication barriers (e.g., aphasia, visual or hearing impairment), and others. MoCAs were performed by occupational therapists blinded to this study as close as possible to the day of discharge as part of standard of care. MoCAs were conducted on patients who were not on sedatives at the time of administration. Hospitalized patients screened from July 2018-May 2019 with no current or previously documented neurological conditions served as controls.

Study Design

MoCAs for hospitalized controls were obtained prospectively as part of a local quality improvement initiative. Sociodemographic characteristics included age, sex, race, insurance type, and highest level of education. Clinical variables and pertinent complications for SAH patients were collected from the chart. These included location of aneurysm (anterior vs. posterior circulation), initial severity of bleed or clinical markers of EBI (Hunt-Hess score, modified Fisher scale, and GCS at time of admission), aneurysm securement procedure (endovascular coiling vs. microsurgical clipping), cerebral vasospasm, DND, and need for a ventriculoperitoneal shunt (VPS) secondary to hydrocephalus. Cerebral vasospasm was determined via daily Transcranial Doppler ultrasound findings or cerebral angiography using previously described criteria or the impression of a neurointerventionalist.¹⁹ DND was defined as the development of delayed focal neurological deficits or decreased level of consciousness with a consistent drop in GCS by ≥ 2 points, with or without the presence of angiographic vasospasm.¹⁶ Our primary outcome was cognition, assessed by MoCA administered prior to hospital discharge. There is lack of consensus in the literature on the normality cutoff score after stroke.^20–22 A cutoff of 26 has been shown to be more sensitive; however, more recent studies have adopted lower cutoffs due to higher specificity. ^7,22,23 Also, using cognitive batteries as a reference, it has been shown that the optimal cutoff for cognitive impairment ranges from 19–22 in the acute stroke phase and from 20–27 in the chronic phase of stroke.²⁰ Thus, in our study, patients were stratified as good vs. poor cognitive status based on a pre-defined MoCA cutoff of 22, where those with a MoCA <22 were considered to have severe cognitive impairment.²⁴ In secondary analysis, we used a MoCA cutoff of 26 or below to indicate cognitive impairment. Our secondary outcome was functional status at time of discharge, assessed via the mRS documented as routine standard of care. Good functional outcome was defined as a mRS of 0–2 while poor outcome was defined as mRS of 3–6.²⁵

Statistical Analysis

Descriptive variables are presented as either percentages, mean and standard deviation (SD), or median and interquartile range (IQR). For initial exploratory analysis, Fisher Exact test or Pearson Chi Square test were used to compare categorical variables while Mann-Whitney test was used to compare continuous variables. Logistic regression analysis was used in SAH patients to identify predictors of cognitive impairment at the time of hospital discharge. Two different multivariate models were used to predict cognitive outcome.²⁶ In model 1, predictors of cognitive outcome with p<0.05 in univariate analysis were adjusted by age and highest level of education, two factors which are known to influence performance on the MoCA screening test.²⁷ In model 2, the results were adjusted by variables with p<0.05 from univariate analysis. The presence of multicollinearity among independent variables was assessed using weighted linear regression and defined as a variance inflation factor ≥5 or tolerance of <0.20.²⁶ Spearman correlation was used to compare cognitive and functional outcomes (MoCA and mRS). A p-value of <0.05 was considered statistically significant for all analyses. Statistical Package for the Social Sciences (Version 25, IBM® SPSS®, Chicago, IL, USA) software was used to conduct the analysis.

RESULTS

A total of 256 aneurysmal SAH patients were screened and 105 met the inclusion criteria (41.0%, Fig. 1). Of those that did not meet inclusion criteria, reasons for no MoCA included: death (15.2%), acute medical status or cognitive deficits exceeding the limit of the MoCA tool (62.3%), patient refusal (3.3%), language barrier compromising the validity of the assessment (8.6%), or other reasons (10.6%). A total of 80 hospitalized control patients were then recruited. Baseline characteristics of control and SAH patients are shown in Table 1. Average length of hospitalization for SAH patients was 15.9 ± 6.1 days. Compared to controls, there was no significance difference by age (p=0.090), sex (p=0.091), insurance status (p=0.765), or highest level of education (p=0.096). However, controls were more likely to be black (62.5% vs. 37.1%; p=0.004). For controls, those with older age were more likely to have cognitive impairment (p=0.033), although this effect was not observed in SAH patients (p=0.438). For SAH patients, there was a difference in highest level of education based on MoCA score (p=0.033).

Figure 1.

Retrospective study design with inclusion and exclusion criteria for SAH patients.

Table 1.

Baseline Characteristics of Control and SAH Patients

	Hospitalized Controls	Total SAH	p-value Control	MoCA >22	Control	MoCA ≥22 p-value SAH	MoCA >22	SAH	MoCA ≥22 p-value
Participants, n	80	105	-	23	57	-	59	46	-
Age, mean (SD)	53.7 (8.3)	51.6 (11.9)	0.090	56.9 (5.8)	52.5 (8.9)	0.033	52.7 (12.4)	50.2 (11.2)	0.438
Female Sex, n (%)	46 (57.5%)	73 (69.5%)	0.091	12 (52.2%)	34 (59.6%)	0.620	43 (72.9%)	30 (65.2%)	0.522
Race, n (%)			0.004			0.420			0.082
Black	50 (62.5%)	39 (37.1%)		16 (69.6%)	34 (59.6%)		27 (45.8%)	12 (26.1%)
White	14 (17.5%)	24 (22.9%)		2 (8.7%)	12 (21.1%)		10 (16.9%)	14 (30.4%)
Other	16 (20.0%)	42 (40.0%)		5 (21.7%)	11 (19.3%)		22 (37.3%)	20 (43.5%)
Insurance, n (%)			0.765			0.350			0.225
Insured	73 (91.3%)	96 (91.4%)		20 (87.0%)	53 (93.0%)		52 (88.1%)	44 (95.7%)
No insurance	6 (7.5%)	6 (5.7%)		3 (13.0%)	3 (5.3%)		5 (8.5%)	1 (2.2%)
Unable to Assess	1 (1.3%)	3 (2.9%)		0 (0.0%)	1 (1.8%)		2 (3.4%)	1 (2.2%)
Highest Level of Education, n (%)			0.096			0.063			0.033
Grade School	6 (7.5%)	6 (5.7%)		2 (8.7%)	4 (7.0%)		2 (3.4%)	4 (8.7%)
High School/G.E.D./Some College	57 (71.2%)	49 (46.7%)		20 (87.0%)	37 (64.9%)		33 (55.9%)	16 (34.8%)
College/Post-College	17 (21.3%)	32 (30.5%)		1 (4.3%)	16 (28.1%)		13 (22.0%)	19 (41.3%)
Unable to Assess	-	18 (17.1%)		-	-		11 (18.6%)	7 (15.2%)

The average MoCA score of SAH patients was 19.9 ± 5.4 while that of controls was 23.3 ± 3.7 (p<0.0001). Approximately 56.2% of SAH patients and 28.8% of controls had a MoCA score of <22 (Fig. 2).

Figure 2.

SAH patients exhibit significant cognitive impairment compared to non-neurological hospitalized controls. (A) Average MoCA score for SAH patients was significantly higher than controls (19.9 ± 5.4 vs. 23.3 ± 3.7, p<0.0001) (B) Percentage of patients with MoCA <22 or ≥22 in SAH and control groups. Cognitive impairment was observed in 56.6% of SAH patients compared to 28.7% of hospitalized controls (p<0.0001).

Further information on SAH patients dichotomized based on MoCA score (<22 vs. ≥22) are provided in Table 2. All patients underwent daily TCDs for assessment of vasospasm and 70.5% of patients underwent follow-up cerebral angiography between days 3–14 after SAH or as clinically indicated. Cerebral vasospasm developed in 66.7% of patients and DND developed in 26.7% of patients. Patients with MoCA <22 were more likely to have high modified Fisher scale (p=0.006) at admission and to develop DND (p=0.026). Hunt-Hess score (p=0.128) and baseline GCS (p=0.068) were comparable in both groups. In univariate analysis, modified Fisher scale (OR 1.801, 95% CI 1.131–2.869; p=0.013) and development of DND (OR 3.079, 95% CI 1.173–8.082, p=0.022) were associated with poor cognitive outcome (Table 3). Notably, while development of DND was associated with poor cognitive outcome, the presence of cerebral vasospasm was not (p=0.572). Multicollinearity was not observed between the independent variables and cognitive outcome. In model 1 of multivariate analysis, modified Fisher scale was a significant independent predictor of cognitive impairment when adjusted for age and highest level of education with an odds ratio of 1.730 (95% CI 1.052–2.843, p=0.031, Table 4). Modified Fisher scale was also an independent predictor of cognitive impairment in model 2 when adjusted by presence of DND (OR 1.638, 95% CI 1.016–2.640, p=0.043). Using a MoCA cutoff of 26 in our secondary analysis, Fisher score remained an independent predictor of cognitive impairment (OR 2.148, 95% CI 1.057–4.366, p=0.035) after adjusting for potential confounding variables but DND did not (OR 3.410, 95% CI 0.731–15.905, p=0.118).

Table 2.

Hemorrhage severity, location, treatment, and delayed sequelae after aneurysmal SAH.

	Total SAH	SAH MoCA >22	SAH MoCA ≥22	p-value
Participants, n	105	59	46
Modified Fisher Scale, median (IQR)	3 (3–4)	3 (3–4)	3 (2–4)	0.006
Hunt-Hess Score, median (IQR)	2 (2–3)	2 (2–3)	2 (2–3)	0.128
Baseline Glasgow Coma Scale, median (IQR)	15 (14–15)	14.5 (14–15)	15 (14–15)	0.068
Aneurysm Location, n (%)				0.771
Anterior Circulation	92 (87.6%)	51 (86.4%)	41 (89.1%)
Posterior Circulation	13 (12.4%)	8 (13.6%)	5 (10.9%)
Aneurysm Treatment, n (%)				0.827
Surgical Clipping	73 (69.5%)	42 (71.2%)	31 (67.4%)
Endovascular Coiling	30 (28.6%)	16 (27.1%)	14 (30.4%)
No Treatment	2 (1.9%)	1 (1.7%)	1 (2.2%)
Ventriculoperitoneal Shunt, n (%)	17 (16.2%)	12 (20.3%)	5 (10.9%)	0.286
Cerebral Vasospasm, n (%)	68 (66.7%)	40 (67.8%)	28 (60.9%)	0.672
Delayed Neurological Deterioration, n (%)	28 (26.7%)	21 (35.6%)	7 (15.2%)	0.026

Table 3.

Univariate Analysis for Predictors of Severe Cognitive Impairment in SAH Patients.

	Univariate
Variable	OR (95% CI)	p-value
Age	1.017 (0.984–1.052)	0.304
Female Gender	1.433 (0.622–3.304)	0.398
Race	0.712 (0.456–1.113)	0.136
Insurance	0.236 (0.027–2.100)	0.196
Highest Level of Education	0.635 (0.305–1.321)	0.224
Modified Fisher Scale	1.801 (1.131–2.869)	0.013
Hunt-Hess Score	1.399 (0.815–2.399)	0.223
Baseline Glasgow Coma Scale	0.968 (0.811–1.154)	0.715
Aneurysm Location	0.777 (0.236–2.557)	0.679
Aneurysm Treatment	1.185 (0.505–2.786)	0.696
Ventriculoperitoneal Shunt	2.094 (0.680–6.444)	0.198
Cerebral Vasospasm	1.270 (0.554–2.908)	0.572
Delayed Neurological Deterioration	3.079 (1.173–8.082)	0.022

Table 4.

Multivariate Logistic Regression for Predictors of Severe Cognitive Impairment.

Variable	Model 1a		Model 2b
	OR (95% CI)	p-value	OR (95% CI)	p-value
Modified Fisher Scale	1.730 (1.052-2.843)	0.031	1.638 (1.016-2.640)	0.043
Delayed Neurological Deterioration	2.106 (0.751-5.909)	0.157	2.287 (0.839-6.238)	0.106

^aIn Model 1, variables with p<0.05 in univariate analysis were each adjusted by age and education, two factors known to influence MoCA performance.

^bIn Model 2, variables with p<0.05 in univariate analysis were adjusted by each other.

We last sought to explore the relationship between cognitive and functional outcome using MoCA and mRS, respectively. Mean mRS at discharge was 1.8 ± 1.3. Good functional outcome was identified in 76/105 (72.4%) of patients while good functional outcome was identified in 29/105 (27.6%) of patients. Patients with a MoCA score <22 had a median mRS of 2 (IQR 1–3), which was significantly higher than those with a MoCA score ≥22 who had a median mRS of 1 (IQR 0–2, p=0.0022) (Figure 3A). There was an inverse correlation between MoCA and mRS (r = −0.3299, p=0.0006): as mRS increased and functional outcome worsened, there was a corresponding increase in the number of SAH patients experiencing significant cognitive impairment (Figure 3B). Of patients with poor functional outcome, 75.9% also exhibited cognitive impairment (Figure 3C). Notably, 48.7% of SAH patients with a good functional outcome (mRS of 0–2) experienced cognitive impairment.

Figure 3.

Patients with a good functional outcome still experience significant cognitive impairment. (A) Median modified Rankin scale (mRS) of cognitively impaired patients was significantly higher in SAH patients with MoCA <22 compared to those with MoCA ≥22 (2 vs. 1, p=0.002). (B) MoCA score showed an inverse correlation with mRS suggesting that as functional outcome worsens, so does cognitive outcome (p<0.0001, r=−0.3410). (C) Percentage of patients with MoCA <22 or ≥22 based on functional outcome shows that 76.7% of patients with poor neurologic outcome while 48.7% of patients with good neurologic outcome have cognitive impairment (p=0.010).

DISCUSSION

Cognitive impairment is highly prevalent after SAH, even in patients with good functional outcome, and can be predicted by a higher modified Fisher scale on admission. In our study, 56.2% of SAH patients exhibited severe cognitive impairment compared to 28.7% of controls. These rates in SAH patients are similar to those reported in the literature.^4,5 Indeed, recent studies have shown that cognitive impairment measured in the subacute stage after SAH largely influences ability of patients to return to work.²⁸ Such high rates of cognitive impairment significantly impact quality of life and productivity in SAH survivors. There was a higher than expected degree of cognitive impairment in hospitalized controls; however, on average this did not meet the threshold for severe cognitive impairment (MoCA <22).

While Hunt-Hess, modified Fisher, and Glasgow Coma scales are all used as markers of SAH severity on admission for prognostic purposes, only the modified Fisher scale predicted cognitive impairment in our study. Hunt-Hess and GCS are widely used clinical scales that assess the degree of brain injury based on the signs and symptoms of the patient at the time of presentation.²⁹ The modified Fisher scale is based on the thickness and location of subarachnoid blood present on a CT scan and has been used to predict the development of vasospasm and DND.^30,31 To our knowledge, this is the first study to identify a role for the modified Fisher scale in predicting cognitive impairment, even after adjustment for age, education, or occurrence of DND. As blood products have been shown to be toxic to neuronal and glial cell populations through inflammatory and pro-apoptotic mechanisms,^32,33 the increased presence of blood and blood products within the subarachnoid space may directly or indirectly contribute to cognitive impairment. Additionally, it has been proposed that increased blood in the subarachnoid space may impair the circulation of cerebrospinal fluid as well as clearance of toxic metabolites and debris through meningeal lymphatics and glymphatic systems.^34,35 However, this area of study warrants further investigation.

Other factors that have been shown to be predictive of cognitive impairment have included age, vasospasm, DND, or cerebral infarction. In our study, we hypothesized that patients who develop cerebral vasospasm and DND would exhibit significant cognitive impairment. We show that age and development of cerebral vasospasm each had no effect on cognitive outcomes. The latter is in line with increasing evidence suggesting that vasospasm alone may not be the sole nor primary factor contributing to outcome after SAH.^8,9,13–15 In our study, DND was associated with cognitive impairment on univariate analysis; however, when adjusted for age, highest level of education, or modified Fisher scale, DND did not reach the level of statistical significance. Despite the inability of our study to identify DND as an independent predictor of cognitive outcome, additional prospective studies in larger patient cohorts are underway which may better address this relationship.¹⁸

Our data also shows a surprisingly high rate of cognitive impairment in SAH patients regardless of their functional outcome. Most clinical trials in SAH to date have focused on functional outcome using scales such as mRS or Glasgow Outcome Scale (GOS) which are heavily influenced by motor deficits, coordination, and balance.^36,37 However, relatively few have incorporated cognitive outcomes as primary endpoints. Our data shows that up to 48.7% of patients with a good functional outcome still experience significant cognitive impairment despite lack of significant neurologic disability. Therefore, clinical trials relying primarily on functional outcome endpoints may be missing the cognitive impairment found in this population. This highlights the need to adopt cognitive outcome endpoints across SAH clinical trials in order to have a more comprehensive assessment of the efficacy of novel therapeutic interventions. There is currently a trial underway using low-dose intravenous heparin after SAH that is assessing 3-month cognitive outcome using MoCA and preliminary results have suggested improved outcomes.³⁸ In addition to assessments for functional outcome, routine screening for neurocognitive decline should be considered an essential component of clinical practice, even in individuals who are deemed to be functionally independent. Our findings also validate the importance of an interdisciplinary team (e.g. occupational, physical and speech therapy, and neuropsychology) in the care of SAH patients in order to fully characterize a patient’s level of disability.

Strengths of our study include incorporation of a larger number of patients compared to previous studies using MoCA to detect cognitive impairment after SAH (Table 5).^7,17,39,40 SAH patients were cared for at a high-volume comprehensive stroke center with a dedicated Neuroscience Intensive Care Unit (ICU) with standardized treatment regimens across patients. We chose the MoCA, which emphasizes tasks involving frontal executive function, attention, and higher-level skills in language, memory, and visuospatial processing that may be important for picking up subtle differences across SAH patients.^7,41 The MoCA is relatively easy to administer and reliable in the ICU setting, is time-efficient, and may be more sensitive for cognitive impairment compared to other questionnaires such as the Mini-Mental Status Examination (MMSE).^42,43 Further, recent common data elements for SAH have recommended the MoCA as a first-line neuropsychological screening test.⁴⁴ A multicenter consortium has also recommended performing MoCA in all SAH patients between 14–28 days of admission.⁴⁵ Once a patient is identified to have significant cognitive impairment based on this screening test, they can be further recommended for follow-up with more comprehensive neuropsychological testing. Eagles et. al (2019) showed that DCI was able to predict cognitive impairment on multivariate analysis using the MMSE and that even those with low mRS scores still showed cognitive impairment; however, they suggest that the MoCA may be a better screening assessment in SAH patients.⁴² Our study expands upon their findings using MoCA and identifies the modified Fisher scale as an additional predictor of cognitive impairment.

Table 5.

Previous Studies Assessing Cognitive Impairment after SAH.

Study	Number of Patients	Time after SAH	Neuropsychological Assessment(s) Performed	Predictors of Cognitive Impairment
Kreiter KT, et al. Stroke. 2002 [4]	113	3 months	Battery of neuropsychological tests covering global mental status, visual memory, verbal memory, reaction time, motor function, executive function, visuo-spatial function, and language function.	Age ≥50, global cerebral edema, left-sided infarction, Hunt-Hess score <2, thick SAH filling anterior interhemispheric fissure, nonposterior circulation aneurysm location
Springer MV, et al. Neurosurgery. 2009 [47]	232	3 months and 1 year	Telephone Interview for Cognitive Status (TICS, cutoff <30)	Fever, anemia, delayed cerebral ischemia (DCI)
Wong GK, et al. J Neurol Neurosurg Psychiatry. 2012 [7]	90	3 months	MoCA (<26) and Mini-Mental Status Examination (MMSE, <27)	Cerebral infarction due to DCI
Schweizer TA, et al. J Neurol Sci. 2012 [17]	32	6 months	MoCA (<26) and MMSE (<27)	Poor World Federation of Neurologic Surgeons (WFNS) grade correlated with lower scores on orientation subtest of MoCA
Wong GK, et al. PLoS One. 2013 [40]	74 (early), 80 (late)	2–4 weeks (early) and 1 year (late)	MoCA (≤18 for early, ≤22 for late)	Did not analyze for predictors
Wong GK, et al. J Neurol Neurosurg Psychiatry. 2013 [48]	120	1 year	Battery of neuropsychological tests including verbal memory, visuospatial skill and memory, attention and working memory, executive function and psychomotor speed, and language domain	Age, DCI
Wong GK, et al. Transl Stroke Res. 2014 [49]	108	2–4 weeks	MoCA (<18)	Did not analyze for predictors
Boerboom W, et al. J Rehabil	59 aSAH, perimesencephalic	1 year	Trail-Making Test (Parts A and B)	Did not analyze for predictors
Med. 2014 [6]	SAH
Wong GK, et al. Eur J Neurol. 2014 [39]	194	1 year	MoCA, no defined cutoff	Did not analyze for predictors
Wong GK, et al. Stroke. 2015 [50]	126	3 months	MoCA and MMSE	DCI, specifically perforator and/or cortical middle cerebral artery location of ischemia
Wallmark S, et al. Acta Neurochir (Wien). 2016 [51]	81	6 months	MoCA (<27)	Age, DCI
James RF, et al. J Neurosurg. 2018 [38]	47	3 months	MoCA (≤20)	Anterior communicating artery aneurysm location, multiple fevers were associated with lower MoCA scores. Treatment with low-dose intravenous heparin associated with higher MoCA scores.
Shen Y, et al. World Neurosurg. 2018 [52]	152	6 months	MoCA (<26)	DCI, hydrocephalus, anterior aneurysm location
Eagles ME, et al. World Neurosurg. 2019 [42]	337	3 months	MMSE (<27)	Poor WFNS grade (IV/V), DCI
Haug Nordenmark T, et al. Acta Neurochir (Wien). 2019 [53]	51	Day of Discharge (median 11, range 2–45 days)	Comprehensive, established a global cognitive impairment index (GCII)	DCI, acute hydrocephalus (requiring CSF drainage >2000 ml)
Rautalin IM, et al. Neurol Sci. 2019 [54]	42	Day of Discharge (mean 23.6 ± 10.2 days)	MoCA (<26) and/or Rapid Evaluation of Cognitive Function (ERFC, <46)	Did not analyze for predictors

There were also several limitations to our study. First, our analysis was conducted at a single institution and despite a relatively high volume of SAH patients, MoCA scores were only available within a subset of these patients (Figure 1). Excluding patients that were unable to complete the MoCA assessment is likely to result in the selection of a healthier population and may underestimate the real impact of SAH on cognition. This is an inherent limitation to all studies addressing cognitive decline in patients in the neurocritical care setting such as those with severe strokes. Further, most other studies have included a smaller sample size or did not analyze for predictors of cognitive outcome (Table 5). The CONSCIOUS-1 trial, in particular, investigated the effect of an endothelin-1 inhibitor on SAH-associated vasospasm. In a post-hoc analysis of this study, Fisher grade did not predict cognitive outcome determined via the MMSE.⁴² However, there is preclinical data suggesting endothelin-1 modulation may affect cognitive impairment,⁴⁶ and as such this prevents a direct comparison between CONSCIOUS-1 and our study. Given the median Hunt-Hess score of 2 in the SAH patients in our study, our conclusions are limited to those with low-grade SAH and should not be generalized to those with high-grade SAH. Therefore, the prevalence of cognitive impairment reported herein may underestimate the true prevalence across all SAH patients. Second, our results are only applicable to the chosen MoCA cutoff of 22, which affords lower sensitivity but higher specificity based on recent literature.^7,22,23 However, cutoff scores for MoCA in the literature have ranged from 18–27 out of 30, with several groups using lower MoCA scores as thresholds for serious cognitive impairment (Table 5). Importantly, the association of cognitive outcome and Fisher score in our study remained statistically significant even after using the more sensitive MoCA cutoff score of 26. By design, our study should be considered a conservative estimate of this association. Third, our study was retrospective in nature and future prospective studies are needed to reproduce these findings in a more standardized manner. The MoCA-Delayed Cerebral Ischemia (DCI) study currently underway in Switzerland aims to assess cognition in a large cohort of patients at standardized time points and would likely expand upon the findings reported here.¹⁸ Additional, more comprehensive neuropsychological assessments could also be performed on patients who screen positive for cognitive impairment on the MoCA.

SUMMARY

In summary, our data highlights an underappreciated degree of cognitive impairment in low-grade SAH patients, many of whom otherwise display good functional outcomes. Higher modified Fisher scale on admission was predictive of which patients may be at higher risk of developing cognitive impairment by the time of discharge. Cognitive screening through routine neuropsychological assessments in low- and high-grade SAH patients should be adopted into clinical practice alongside more traditional functional outcome measures.