Juntendo Medical Journal, 2026 Original Articles Juntendo Medical Journal 2026 Early Clinical, Laboratory, and Imaging Correlates of Neurological Dysfunction in Adults Presenting to the Emergency Department with Sepsis: A Single-Center Retrospective Study Florin SCARLATESCU1, 2), Ecaterina SCARLATESCU2, 3), Daniela BARTOS2) 1)Department of Neurology, Clinical Emergency Hospital, Bucharest, Romania 2)University of Medicine and Pharmacy Carol Davila, Bucharest, Romania 3)Department of Anaesthesia and Intensive Care, Fundeni Clinical Institute, Bucharest, Romania Objectives: To characterize the prevalence and clinical profile of neurological dysfunction and pure sepsis-associated encepha­ lopathy (SAE) in adults presenting to the emergency department (ED) with sepsis, and to identify early clinical, laboratory, and brain imaging parameters associated with these phenotypes. Materials: This retrospective study included 226 adults with sepsis (Sepsis-3 criteria) at a tertiary emergency hospital. Neuro­ logical dysfunction was defined as an acute mental status change, and pure SAE was diagnosed after systematic exclusion of other identifiable causes of encephalopathy. Demographic, clinical, laboratory, and brain computed tomography (CT) data were collected from medical records. Methods: Group comparisons were made using Mann-Whitney U, chi-square, or Fisher’s exact tests. Multivariable binary logistic regression identified predictors of neurological dysfunction and pure SAE. Results: Neurological dysfunction was present in 79.6% of patients, and pure SAE in 24.8%. In the multivariate model, red cell distribution width (RDW) independently predicted pure SAE (OR 1.20, 95% CI 1.05-1.36; p = 0.005), while creatinine was inversely associated (OR 0.77, 95% CI 0.60-0.99; p = 0.038), consistent with the exclusion of patients with metabolic causes of encephalopathy from the pure SAE category. Frontal cortical atrophy on brain CT was the only imaging parameter linked to pure SAE (p = 0.022). Neurological dysfunction and pure SAE were associated with lower survival at discharge (26% vs. 67.3% and 19.6% vs. 39.4%, respectively). Conclusions: Neurological dysfunction is common in ED sepsis patients. RDW and frontal cortical atrophy may help identify pure SAE early. These markers could aid rapid risk stratification and management. Key words: sepsis-associated encephalopathy, neurological dysfunction, red cell distribution width, brain CT, cortical atrophy Introduction Sepsis-associated encephalopathy (SAE) refers to widespread neurological dysfunction resulting from an abnormal systemic response to severe infections or sepsis, occurring in the absence of direct central nervous system infection, acute brain injury, or alternative causes of encephalopathy1). SAE presents with a broad spectrum of neurolog­ ical symptoms, from mild confusion, hypersomno­ lence, and delirium to more severe states such as stupor or coma. This heterogeneity, coupled with the lack of specific biomarkers, contributes to freq­ uent underrecognition and diagnostic uncertainty Corresponding author: Ecaterina Scarlatescu Department of Anaesthesia and Intensive Care, Fundeni Clinical Institute 258 Fundeni Street, 022328 Bucharest, Romania TEL: +40-213180444 〔Received E-mail: katyscarlatescu@yahoo.com Mar. 7, 2026〕 〔Accepted Mar. 12, 2026〕 J-STAGE Advance published date: Apr. 10, 2026 Copyright © 2026 The Juntendo Medical Society. This is an open access article distributed under the terms of Creative Commons Attribution License (CC BY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original source is properly credited. doi: 10.14789/ejmj.JMJ26-0022-OA 1 Scarlatescu F, et al: Early correlates of sepsis-associated encephalopathy in the emergency department in routine clinical practice2). Sepsis-associated encephalopathy (SAE) can be classified into two types: an early-onset form, which often precedes other organ dysfunction or overt signs of infection, and a later-onset form. The later type is multifactorial, resulting not only from sepsis itself, but also from factors such as multiple organ failure, sedative medications, and the conditions asso­ ciated with intensive care3). Diagnosing early-onset SAE is particularly challenging, because neurolog­ ical dysfunction may be the sole or leading manifes­ tation at the time of evaluation in the emergency department (ED), and the clinical picture—usually an acute alteration in mental status without focal deficits or meningeal signs—is highly non‑specific. In this setting, SAE must be distinguished from a wide array of alternative diagnoses, including meta­ bolic, toxic, structural, and primary neurological disorders, and a high index of suspicion is required to ensure timely recognition and appropriate mana­ gement2). The reported prevalence of brain dysfunction among patients with sepsis reaches up to 70%, although estimates vary considerably across studies due to differences in inclusion criteria4, 5). In a recent investigation, sepsis-associated encephalopathy (SAE) was identified in 411 out of 603 patients admitted to the intensive care unit (ICU) for sepsis secondary to pulmonary infection6). The majority of available literature pertains to ICU populations, while studies addressing brain dysfunction in pati­ ents at the point of hospital admission—regardless of illness severity or subsequent ICU transfer— remain limited. At present, there are no universally accepted diagnostic criteria for SAE, and diagnosis remains largely one of exclusion. In practice, SAE is estab­ lished after ruling out alternative causes of enceph­ alopathy through clinical evaluation, basic labora­ tory testing, and, where appropriate, neuroimaging and cerebrospinal fluid analysis. In many patients, a thorough neurological examination combined with routine laboratory investigations is sufficient to support the diagnosis. Neuroimaging is generally reserved for a minority of patients, primarily to exclude cerebrovascular events or traumatic brain injury. For instance, a pediatric study reported that 21% of children with sepsis exhibited brain lesions detectable by magnetic resonance imaging 2 (MRI)7). In adult ICU populations with traumatic brain injury, MRI has revealed both new cerebral lesions and the extension of existing injuries in the context of infection and sepsis8). While MRI is commonly utilized in research on sepsis, there remains a paucity of data concerning the diagnostic and prognostic value of brain computed tomog­ raphy (CT) findings at the time of hospital admis­ sion, particularly during the very early phase of sepsis. Despite growing interest in SAE, several impor­ tant gaps remain. First, most studies focus on ICU patients and later time‑points, when neurological status is confounded by sedation, prolonged organ support, and cumulative iatrogenic factors. Second, few investigations distinguish “pure” SAE—defined as encephalopathy primarily attributable to sepsis —from multifactorial encephalopathy driven by concurrent renal, hepatic, or other metabolic deran­ gements. Third, there is limited evidence on simple, bedside‑available clinical, laboratory, and imaging markers that can help clinicians differentiate and predict SAE early, at the time of ED presentation, when diagnostic and therapeutic decisions are most consequential. Against this background, the present study was designed with two main objectives. First, to charac­ terize the prevalence and clinical profile of neuro­ logical dysfunction and pure SAE in adults presen­ ting to the ED with sepsis, independently of subse­ quent ICU admission. Second, to identify early clin­ ical, laboratory, and brain imaging parameters— readily available at or shortly after ED arrival— that are associated with neurological dysfunction at admission and with pure SAE, and to develop multivariable models capable of predicting these phenotypes early in the hospital course. By focusing on the ED setting and explicitly separating pure SAE from multifactorial encephalopathy, this research addresses a critical gap in the current literature and aims to provide clinicians with practical tools to recognize SAE promptly, refine etiological attri­ bution of altered mental status, and potentially guide risk stratification and early management in routine clinical practice. Materials and Methods Patient selection This retrospective observational study enrolled Juntendo Medical Journal, 2026 patients with sepsis, as defined by the Sepsis-3 criteria9), following confirmation of diagnosis in the emergency department. Data were collected from Clinical Emergency Hospital Bucharest, a highvolume tertiary care center in Romania. The study protocol received ethical approval from the institu­ tional ethics committee (approval number 2257/22. 01.2024), which waived informed consent require­ ments owing to the retrospective, non-interven­ tional design and reliance on routinely collected clinical data. In accordance with institutional policy, all patients provide written consent for the use of clinical data obtained during their hospital admis­ sion. The diagnosis of neurological dysfunction at admission and pure SAE Neurological dysfunction at admission was defined as an acute alteration in mental status—including confusion, disorientation, a reduced level of conscio­ usness, agitation or delirium—detected during the initial clinical evaluation in the ED. The diagnoses of neurological dysfunction at admission and pure SAE were established by the neurologist examining the patients. A diagnosis of pure SAE was estab­ lished after systematic exclusion of other identifi­ able causes of neurological impairment, including direct central nervous system infection (menin­ gitis, encephalitis, brain abscess), acute cerebrovas­ cular events, traumatic brain injury, severe hepatic encephalopathy in patients with cirrhosis, severe uremic encephalopathy in patients with advanced renal failure, primary metabolic derangements (e.g. hypoglycemia, severe electrolyte disturbances) judged to be the dominant cause, drug-related causes (sedation, intoxication), and pre-existing neuropsychiatric conditions that could fully account for the clinical picture. In patients where encepha­ lopathy was considered multifactorial—i.e. driven by a combination of sepsis and concurrent organ failures such as uremia or hepatic insufficiency— the episode was not classified as pure SAE. Demographic and clinical data were retrieved from the electronic medical records for all enrolled patients and included age, sex, pre-existing comor­ bidities (arterial hypertension, atrial fibrillation, heart failure, diabetes mellitus, chronic kidney disease, cirrhosis, active cancer and metastatic cancer), the source and microbiological profile of infection, length of stay in the intensive care unit (ICU) and in the hospital, and survival at hospital discharge. Blood tests The blood samples were obtained from all patients in the ED at the time of initial presentation, before initiation of fluid resuscitation or antibiotic therapy in most cases. The following parameters were reco­ rded: white blood cell count (WBC), hemoglobin (Hb), mean corpuscular volume (MCV), platelet count (PLT), platelet distribution width (PDW), red cell distribution width (RDW), mean platelet volume (MPV), glucose, serum creatinine, blood urea, serum sodium, and international normalized ratio (INR). All analyses were performed using the hospital's central laboratory, with automated hema­ tology and biochemistry analyzers calibrated according to standard institutional protocols. Imaging Brain computed tomography (CT) was performed during the initial hospital admission in a subset of patients, according to clinical indication (e.g. acute alteration of consciousness, focal neurological signs, or clinical suspicion of intracranial pathology). CT imaging was the method used to rule out acute intracerebral lesions, but in the same time the CT findings were explored to identify simple imaging markers associated with SAE. All available CT scans were retrospectively reviewed by a single investigator (neurologist, SF), and the following structural brain parameters were systematically recorded: total number of cerebral lacunes and macroinfarcts; third ventricle diameter (measured in millimeters at the widest point on axial images); cortical atrophy: graded separately for each lobe (frontal, temporal, parietal, occipital) and as a global cortical atrophy (GCA) score using a 4-point visual rating scale (0 = no atrophy, 1 = mild, 2 = moderate, 3 = severe)10); white matter disease: graded using the Fazekas scale (0 = absent, 1 = punctate foci, 2 = beginning confluent, 3 = confl­ uent periventricular white matter hyperintensi­ ties)11, 12); medial temporal atrophy (MTA): scored according to the Scheltens visual rating scale (0-4), where 0 indicates no atrophy and 4 indicates severe medial temporal lobe atrophy with marked widening of the choroid fissure and temporal horn13, 14); ventri­ cular indices as markers of brain atrophy: the bicau­ 3 Scarlatescu F, et al: Early correlates of sepsis-associated encephalopathy in the emergency department date index (ratio of the minimum intercaudate distance to the brain width at the same level) and the Evans index (ratio of the maximum width of the frontal horns to the maximum internal diam­ eter of the skull) were calculated on axial images as quantitative markers of central brain atrophy and ventricular enlargement15-17). Statistical analysis Continuous variables were tested for normality using the Shapiro-Wilk test. Data with normal distribution were reported as mean ± SD. As the majority of laboratory and imaging parameters showed non-normal distributions, continuous and ordinal data were expressed as median (interquar­ tile range) and compared between groups using the Mann-Whitney U test. Categorical and binary variables were expressed as absolute frequencies and percentages and compared using the chi-square test or Fisher's exact test, as appropriate. For the identification of independent predictors of neuro­ logical dysfunction at admission and pure sepsisassociated encephalopathy, variables with a p-value below 0.10 in the univariate analysis were entered into multivariable binary logistic regression models. Results of the logistic regression are reported as adjusted odds ratios (OR) with 95% confidence intervals (CI). Multicollinearity among predictors entered in the logistic regression model was assessed in SPSS by running a separate linear regression with these variables entered as independent predic­ tors and requesting collinearity diagnostics (toler­ ance, variance inflation factor, and collinearity diag­ nostics table). For this study, a p value of < 0.05 was considered significant. Statistical analyses were performed using SPSS Statistics v 23.0 (IBM Corp). Results A total of 226 patients (42.9% men, 57.1% women) with sepsis were included in the analysis. The mean (± SD) age was 72.69 (± 15.07) years. Overall neuro­ logical dysfunction at admission was very common, affecting nearly 80% of patients (n = 180). Pure SAE was identified in approximately one quarter of patients (n = 56, 24.8%). Baseline characteristics Patients with and without neurological dysfunc­ tion at admission had similar age (mean age 72.54 4 vs 73.26 years). Patients with pure SAE were slightly younger than those without (mean age ≈70.5 vs 73.3 years), although the difference was modest. Sex distribution was similar between groups, with men representing 42.2% and 44.6% of the cases with neurological dysfunction at admission and with pure SAE, respectively. Patients with and without neurological dysfunction at admission had no statistically significant differences in hyperten­ sion, atrial fibrillation, heart failure, diabetes, cirrh­ osis, cancer, or metastatic cancer incidence. Chronic kidney disease tended to be more frequent in patients with neurological dysfunction at admission, but this difference also did not reach conventional significance (Table 1). When patients with and without pure SAE were compared, age and most comorbidities again showed no significant variation between groups. However, chronic kidney disease was significantly less common in patients with pure SAE, whereas cirrhosis was observed only in patients without this diagnosis, consistent with the clinical attribution of their asso­ ciated neurological dysfunction to hepatic or multi­ factorial mechanisms (Table 1). Organ dysfunction, laboratory data and outcomes Among patients with neurological dysfunction at admission, hemoglobin, leukocyte and platelet counts were comparable to those without neurological involvement, but red cell distribution width (RDW) and mean platelet volume (MPV) were higher, suggesting differences in hematologic activation. The patients presenting neurological dysfunction at admission also exhibited more pronounced renal dysfunction, with higher creatinine and urea, as well as higher sodium and INR, consistent with broader organ failure. Neurological dysfunction at admission was associated with a longer intensive care stay, while overall hospital length of stay did not differ significantly (Table 2). When stratified by pure SAE, standard hemato­ logic parameters and platelet indices showed no major differences between groups. In contrast, patients with SAE had better preserved renal func­ tion, with lower creatinine and urea values, and a shorter hospital stay compared with those without this diagnosis, whereas ICU length of stay was similar (Table 2). In the multivariate model restricted to laboratory Juntendo Medical Journal, 2026 Table 1 Baseline demographic and comorbidity characteristics of sepsis patients stratified by neurological dysfunction at admission and pure sepsis-associated encephalopathy Neurological dysfunction at admission P value Pure sepsis-associated encephalopathy P value YES (n = 180) NO (n = 46) YES (n = 56) NO (n = 170) Age (years) 72.5(± 15.3) 73.2 (± 14.9) 0.771 70.5 (± 16.7) 73.3 (± 14.4) 0.240 AHT 92 (82.1%) 20 (43.4%) 0.350 26 (46.4%) 86 (50.5%) 0.589 AF 70 (38.9%) 12 (26.1%) 0.10 16 (28.5%) 66 (38.8) 0.166 HF 49 (27.2%) 13 (28.3%) 0.880 14 (25%) 48 (28.2%) 0.638 CKD 58 (32.2%) 9 (19.5%) 0.091 8 (14.2%) 59 (34.7%) 0.004* DM 55 (30.5%) 18 (39.1%) 0.262 16 (28.5%) 57 (33.5%) 0.490 Cirrhosis 8 (4.4%) 2 (4.3%) 0.977 0 (0%) 10 (5.9%) 0.016* Cancer 31 (17.2%) 11 (23.9%) 0.298 14 (25%) 28 (16.4%) 0.155 Metastatic cancer 10 (5.8%) 5 (10.9%) 0.196 5 (8.9%) 10 (5.9%) 0.427 Data are expressed as mean ± SD or n (%). P value was obtained using the t-test or chi-square test, as appropriate. * p statistically significant (< 0.05); AHT, arterial hypertension; AF, atrial fibrillation; HF, heart failure; CKD, chronic kidney disease; DM, diabetes mellitus. Table 2 Laboratory parameters, and clinical outcomes of sepsis patients stratified by neurological dysfunction at admission and pure sepsis-associated encephalopathy Parameter WBC (/µL) Reference ranges 4000-10000 Neurological dysfunction at admission YES (n = 180) P value NO (n = 46) 15000 (12430) 15650 (14980) Pure sepsis-associated encephalopathy YES (n = 56) 0.825 P value NO (n = 170) 13700 (14125) 15700 (12585) 0.648 Hemoglobin (g/dL) 11.7-16 11.3 (4) 12.2 (3) 0.130 11.4 (4) 11.5 (4) 0.339 RDW (%) 11.5-14.5 14.4 (3) 13.3 (2) 0.026* 14.6 (3) 14.2 (3) 0.099 Platelet count (*1000/µL) 150-400 193.5 (140) 222.5 (131) 0.200 214 (118) 194.5 (144) 0.801 PDW (%) 9-17 17.6 (2) 17.3 (2) 0.137 17.4 (2) 17.6 (2) 0.448 MPV (fL) 7-11 8.4 (3) 7.4 (4) 0.034* 7.9 (3) 8.4 (3) 0.887 Creatinine (mg/dL) 0.55-1.02 2 (2) 1.3 (2) 0.001* 1.24 (1) 2.25 (2) <0.001* Urea (mg/dL) 19-49 103 (112) 69.5 (44) 0.001* 58.5 (52) 103 (111) <0.001* Sodium (mmol/L) 135-145 139.9 (12) 138 (10) 0.03* 139 (11) 139 (11) 0.569 INR 0.9-1.2 1.38 (1) 1.21 (0) 0.008* 1.38 (0) 1.31 (1) 0.371 ICU LOS (days) 2 (6) 0 (6) 0.031* 2 (6) 1.5 (6) 0.326 Hospital LOS (days) 5 (12) 8 (8) 0.23 3.5 (8) 7 (12) 0.023* Survival at hospital discharge 47 (26%) 31 (67.3%) <0.001* 11 (19.6%) 67 (39.4%) <0.007* Data are expressed as median (IQR) or n (%). P value was obtained using the Mann Whitney U test or chi-square test, as appropriate. * p statistically significant (< 0.05); WBC, white blood cells; RDW, red blood cells distribution width; PDW, platelet distribution width; LOS, length of stay and hematologic parameters, none of the candidate variables (RDW, MPV, creatinine, urea, sodium, INR) showed an independent association with neurological dysfunction at admission at the conventional significance threshold. The point esti­ mates suggested only modest effects, and confi­ dence intervals were wide and crossed unity, indi­ cating that, once their inter correlations were taken into account, these indices did not reliably distin­ guish patients with and without neurological dysfunction at hospital admission (Table 3). In contrast, the model for pure SAE identified RDW and creatinine as independent predictors. For RDW, the adjusted odds ratio of approximately 1.20 per unit increase (95% CI roughly 1.05-1.36) indicates that for each 1% increment in RDW the odds of sepsis associated encephalopathy increased by about 20%, after controlling for the other labora­ 5 Scarlatescu F, et al: Early correlates of sepsis-associated encephalopathy in the emergency department Table 3  ultivariable logistic regression analysis of independent predictors of neurological dysfunction M at admission and pure sepsis-associated encephalopathy Neurological dysfunction at admission OR (95% CI) P value Sepsis-associated encephalopathy OR (95% CI) P value RDW (%) 1.16 (0.96-1.39) 0.124 1.20 (1.05-1.36) 0.005* MPV (fL) 1.11 (0.93-1.34) 0.255 Creatinine (mg/dL) 1.11 (0.81-1.52) 0.506 0.77 (0.60-0.99) 0.038* Urea (mg/dL) 1.01 (1.00-1.02) 0.06 1.00 (0.99-1.00) 0.123 Sodium (mmol/L) 1.03 (0.99-1.08) 0.139 INR 1.13 (0.74-1.72) 0.574 Predictor RDW, red blood cells distribution width; MPV, mean platelet volume; * p statistically significant (<0.05) tory variables in the model. Creatinine showed an opposite pattern, with an adjusted odds ratio of about 0.77 per 1 mg/dL (95% CI around 0.60-0.99), meaning that each 1 mg/dL increase in creatinine was associated with a 23% reduction in the odds of being classified as having pure SAE. This inverse association supports the interpretation that patients with more severe renal impairment are less often categorized as having pure SAE and are instead considered to have multifactorial or metabolically driven encephalopathy. Urea did not remain inde­ pendently significant once RDW and creatinine were included in the model. Of note, creatinine and urea showed moderate collinearity (Spearman rho = 0.688), with variance inflation factors of approxi­ mately 2 for both variables, which may have contri­ buted to the attenuation of the urea effect in the adjusted model while not invalidating the regres­ sion estimates (Table 3). Brain imaging findings CT scan was performed in 92 patients, and the imagistic findings were compared between patients with and without neurological dysfunction at admi­ ssion and pure SAE. Among the 92 patients with available brain CT, 84 had neurological dysfunction at admission and 8 did not; 26 were classified as having pure SAE and 66 were not. The number of cerebral lacunes, macroinfarcts, and third ventricle diameter were similar between patients with and without neurological dysfunction at admission, with no statistically significant differ­ ences. Likewise, the bicaudate index and Evans index did not differ between groups. Among the cortical atrophy scores, temporal cortical atrophy showed a trend towards more severe grades in 6 patients with neurological dysfunction, approaching conventional significance, while global cortical atrophy showed a weaker, non-significant trend in the same direction. No significant differences were found for frontal, parietal, or occipital cortical atrophy, nor for medial temporal atrophy (MTA Scheltens) or Fazekas scale scores (Table 4). When patients were stratified by pure SAE, the number of lacunes, macroinfarcts, third ventricle diameter, bicaudate index, and Evans index again showed no significant differences between groups. Among the regional cortical atrophy scores, only frontal cortical atrophy demonstrated a significantly different distribution between patients with and without pure SAE. Notably, this difference was driven by a markedly lower proportion of grade 1 (mild) frontal atrophy in patients with SAE compared with those without, alongside a shift towards higher grades of atrophy (grades 2 and 3), suggesting a bimodal distribution in the SAE group rather than a simple linear increase in atrophy severity. Global, temporal, parietal, and occipital cortical atrophy, as well as medial temporal atrophy and Fazekas scale scores, did not differ significantly between the two groups (Table 4) Taken together, these imaging findings indicate that the burden of chronic cerebrovascular and neurodegenerative lesions detectable on routine brain CT was broadly similar between patients with and without neurological dysfunction at admission or pure SAE. The only significant imaging associa­ tion identified was between frontal cortical atrophy and pure SAE, suggesting that pre-existing frontal lobe vulnerability may contribute to the clinical expression of this phenotype, although this finding should be interpreted cautiously given the relatively Juntendo Medical Journal, 2026 Table 4 Brain CT imaging findings in sepsis patients stratified by neurological dysfunction at admission and pure sepsisassociated encephalopathy Neurological dysfunction at admission Parameter P value Pure sepsis-associated encephalopathy P value YES (n = 84) NO (n = 8) YES (n = 26) NO (n = 66) Number of cerebral lacunes 1(3) 0.5 (6) 0.834 1 (3) 1 (3) 0.724 Number of macroinfarcts 0 (1) 0 (0) 0.318 0 (1) 0 (1) 0.854 V3 diameter (mm) 11.3 (4.9) 10.5 (5.3) 0.708 10.5 (7) 11.5 (4.3) 0.706 Bicaudat index 0.2 (0.06) 0.19 (0.1) 0.978 0.19 (0.09) 0.2 (0.06) 0.673 Evans index 0.3 (0.07) 0.32 (0.07) 0.311 0.3 (0.07) 0.3 (0.08) 1 Global cortical atrophy Frontal cortical atrophy (Frontal GCA) Temporal cortical atrophy (Temporal GCA) Parietal cortical atrophy (Parietal GCA) Occipital cortical atrophy (Occipital GCA) MTA Scheltens Fazekas Scale 0 13 2 6 9 1 17 4 4 17 2 44 2 13 33 3 10 0 3 7 0 15 2 7 10 1 22 4 2 24 2 38 2 13 27 3 9 0 4 5 0 15 2 6 11 1 23 5 6 22 2 39 1 11 29 3 7 0 3 4 0.128 0.279 0.093 0 16 3 6 13 1 26 3 7 22 2 38 2 13 27 3 3 0 0 3 0 36 5 10 31 1 32 2 10 24 2 16 1 6 11 0.488 0.567 3 0 0 0 0 0 13 2 6 9 1 21 4 5 20 2 30 1 9 22 3 16 1 5 12 4 4 0 1 3 0.385 0 22 2 8 16 1 33 3 9 27 2 16 2 4 15 3 12 1 5 8 0.988 0.589 0.022* 0.622 0.443 0.694 0.755 0.714 Data are expressed as median (IQR) or n (number of patients). p value was obtained using the Mann Whitney U test or chi-square test, as appropriate. * p statistically significant (< 0.05) small imaging subgroup and the multiple compari­ sons performed. Discussion The present study found that nearly 80% of adults presenting to the ED with sepsis had neurological dysfunction at admission, a prevalence substantially higher than the commonly cited range of 8-70% from earlier ICU-based series4). This difference is likely explained by the inclusive definition adopted 7 Scarlatescu F, et al: Early correlates of sepsis-associated encephalopathy in the emergency department in this study, which captured the full spectrum of altered mental status—from mild confusion and disorientation to coma—at the very first clinical contact, before any therapeutic intervention. Prev­ ious estimates were typically generated in ICU cohorts, at later time-points when sedation, fluid resuscitation, or metabolic correction may have already attenuated or confounded neurological signs. When a strict exclusion-based definition was applied to identify pure SAE—attributing encephalopathy primarily to sepsis after ruling out hepatic, uremic, metabolic, structural, and drug-related causes— the prevalence fell to 24.8%. This figure is consis­ tent with the lower end of reported SAE incidence (15.0-62.4%) in the recent systematic review by Zhou et al. (2025)18), which documented wide vari­ ability across studies depending on diagnostic criteria, patient populations, and care settings. Notably, the strict exclusion of cirrhotic patients (none classi­ fied as pure SAE), patients with severe renal impairment, and those with multifactorial encepha­ lopathy likely accounts for the relatively conserva­ tive prevalence of pure SAE in the present cohort. An important finding of this study is the substan­ tially worse survival in patients with neurological dysfunction at admission (26% vs. 67.3%) and in those with pure SAE (19.6% vs. 39.4%). These figures are in line with the seminal report by Sprung et al. (1990), which first demonstrated that the mortality rate of septic patients with altered mental status was 49% compared with 26% in those with normal mental status19). In a large international post hoc analysis of the ICON audit encompassing 7,192 crit­ ically ill patients, Crippa et al. confirmed that brain failure combined with sepsis was independently associated with a markedly increased risk of hospital death (adjusted RR 5.61, 95% CI 3.93-8.00), substan­ tially higher than either sepsis or brain failure alone20). Sonneville et al. (2017), in a retrospective analysis of 2,513 patients, found that even mild alterations of mental status (GCS 13-14) at ICU admission were independently associated with increased mortality (HR 1.38, 95% CI 1.09-1.38), while severe encephalopathy (GCS 3-8) carried a hazard ratio of 3.37 (95% CI 2.82-4.03)21). A 2026 systematic review of risk prediction models for SAE confirmed that 28-day mortality in SAE can reach up to 46%, and that SAE prevalence ranges from 8% to 70% depending on diagnostic criteria22). 8 Shen et al. (2025) extended these findings by demon­ strating that dynamic GCS trajectories over the first 5 ICU days carry distinct prognostic implica­ tions—patients with persistently high GCS scores had better survival compared to those with decrea­ sing GCS scores23). The current study extends this body of evidence by demonstrating that neurolog­ ical dysfunction, assessed at the earliest clinical contact in the ED—before sedation, fluid resuscita­ tion, or ICU-related confounders—already carries strong prognostic significance regardless of subse­ quent ICU admission. This early-assessment pers­ pective is particularly important because the vast majority of prior studies examined prognosis using neurological evaluations performed after ICU admi­ ssion, when the clinical picture is frequently confo­ unded by iatrogenic factors. Among all laboratory parameters evaluated, RDW emerged as the only independent positive predictor of pure SAE in the multivariate model, with each 1% increment in RDW being associated with a 20% increase in the odds of having pure SAE after controlling for creatinine and urea. This finding adds to a growing body of evidence linking elevated RDW to sepsis severity and organ dysfunc­ tion. A meta-analysis by Zhang et al. (2020), encom­ passing 17,961 sepsis patients across 11 studies, demonstrated that increased baseline RDW was significantly associated with mortality (pooled HR 1.14, 95% CI 1.09-1.20)24). Similarly, Singh et al. demonstrated that RDW had excellent diagnostic accuracy (AUC = 0.823) for predicting sepsis morta­ lity, performing comparably to serum lactate25). The biological plausibility of the RDW-SAE association is well-established. Sepsis induces a state of systemic inflammation with markedly increased oxidative stress, which profoundly affects erythrocyte biology. Beyond reflecting systemic inflammation, elevated RDW may also indicate impaired erythrocyte defor­ mability and microcirculatory dysfunction, which are increasingly recognized as contributors to organ injury in sepsis. Inflammatory cytokines and oxidative stress can disrupt erythrocyte membrane integrity and cytoskeletal stability, resulting in heterogeneous red blood cell populations and reduced deformability26). These alterations may impair capillary transit and oxygen delivery in the micro­ circulation27). Because cerebral perfusion relies on efficient microvascular flow, disturbances in red Juntendo Medical Journal, 2026 blood cell rheology could contribute to regional cerebral hypoxia and metabolic stress in vulner­ able brain regions. Microcirculatory alterations, including endothelial activation and heterogeneous capillary perfusion, are well-recognized features of sepsis and have been implicated in the pathogen­ esis of SAE28). In this context, RDW may represent a readily available surrogate marker of systemic processes affecting cerebral microcirculatory integ­ rity24). However, RDW has also been associated with overall disease severity in critically ill patients. Therefore, the observed association with SAE may partly reflect a greater systemic inflammatory burden, and future studies incorporating estab­ lished severity indicators such as SOFA score or lactate levels will be necessary to clarify its inde­ pendent predictive value. Pro-inflammatory cytokines—including TNF-α, IFN-γ, IL-1β, and IL-6—suppress erythroid proge­ nitor cell activity, impair erythrocyte deformability, and alter RBC membrane structure, resulting in anisocytosis and measurably increased RDW26). Recent in-vitro and in-silico evidence has confirmed that oxidative stress is a direct mechanistic driver of increased RDW29). It is therefore conceivable that the heightened systemic inflammation reflected by elevated RDW also drives the neuroinflamma­ tory cascades underlying SAE, including bloodbrain barrier disruption, microglial activation, and neuronal dysfunction30). From a clinical perspec­ tive, RDW is universally available, inexpensive, and reported as part of a standard complete blood count. Its identification as an independent predictor of pure SAE in this study supports its potential utility as an early bedside marker for SAE risk stratifica­ tion in the ED, where more sophisticated biomarkers (S100β, NSE) or neuroimaging are often not imme­ diately available. A recent systematic review of SAE prediction models noted that while age, SOFA score, and serum sodium were the most frequently included predictors, RDW has not yet been incorporated into published SAE prediction tools, representing a novel and readily implementable addition18). Creatinine showed an inverse association with pure SAE in the multivariate model (OR 0.77, 95% CI 0.60-0.99; p = 0.038), a finding that is counterintu­ itive at first glance but is fully consistent with the study's diagnostic framework. Patients with more severe renal impairment—manifested by higher creatinine and urea levels—were systematically excluded from the pure SAE category if their encephalopathy was judged to be predominantly uremic or multifactorial in origin. Thus, the lower creatinine observed in the pure SAE group does not imply that preserved renal function causes SAE; rather, it reflects the clinical attribution process whereby encephalopathy in patients with advanced renal dysfunction is classified as multifac­ torial rather than as pure sepsis-driven. This inter­ pretation is supported by the univariate findings: creatinine and urea were both significantly higher in patients with neurological dysfunction at admis­ sion overall (p = 0.001 for both), consistent with the well-recognized role of acute kidney injury as a contributor to neurological impairment in sepsis. Sepsis-associated acute kidney injury is itself a highly prevalent complication, and the interplay between renal failure and cerebral dysfunction is complex, involving accumulation of uremic toxins, altered drug clearance, and compounding meta­ bolic derangements31). The present findings under­ score the importance of carefully distinguishing the etiological attribution of encephalopathy in clinical practice, as the management implications of pure SAE versus uremic or hepatic encephalopathy may differ substantially. Regarding the imagistic findings, among the 92 patients who underwent brain CT, frontal cortical atrophy was the only imaging parameter signifi­ cantly associated with pure SAE (p = 0.022). This resonates with accumulating evidence that pre-ex­ isting brain vulnerability—whether from neurode­ generation, cerebrovascular disease, or atrophy— predisposes to delirium and acute brain dysfunction during systemic illness. Pendlebury et al. demon­ strated that white matter changes and cerebral atrophy on brain imaging obtained up to 5 years earlier predicted delirium during subsequent hospital admissions in a stroke/TIA cohort32). The frontal lobes are particularly relevant given their role in attention, executive function, and behavioral regu­ lation—the very cognitive domains most affected in SAE and delirium. Functional MRI studies in sepsis survivors have specifically identified decreased degree centrality and nodal efficiency in frontal regions (particularly the right orbital inferior frontal gyrus), suggesting that the frontal lobe is a key node of vulnerability in sepsis-associated neurolog­ 9 Scarlatescu F, et al: Early correlates of sepsis-associated encephalopathy in the emergency department ical injury33). Nakae et al. reported rapidly progres­ sive brain atrophy in septic ICU patients using semiautomatic CT volumetry, with a mean brain volume decrease of 3.7% over a median of 31 days34). While the present study assessed pre-existing atrophy on a single CT obtained at admission rather than progressive volume loss, the two findings are complementary: patients who already have frontal atrophy may be more susceptible to SAE at presen­ tation, and sepsis may subsequently accelerate further brain volume loss. The present study's approach—focusing on simple, routinely available ED parameters including RDW and brain CT findings—complements existing SAE prediction models, which have predominantly been developed in ICU settings using large administra­ tive databases18). A notable distinction of the present study is its focus on the ED setting, which shifts the clinical perspective from predicting SAE among established ICU patients to identifying neurological dysfunction at the point of first medical contact. This is clinically meaningful because the ED repre­ sents the earliest window of opportunity for recog­ nition and intervention, before sedation, mechanical ventilation, and other ICU-related confounders obscure the neurological picture. Additionally, the explicit separation of pure SAE from multifactorial encephalopathy represents a conceptual refinement that few prior studies have attempted, yet one that has clear clinical implications for diagnostic workup and management35). Recently, Tian et al. devel­ oped a predictive model for SAE based on immuno­ logical markers (IFN-γ, TNF-α, and CD4+/CD8+ ratio), demonstrating the potential of immune biomarkers in SAE prediction36). While such markers may offer superior pathophysiological specificity, they are not routinely available in most EDs. The identification of RDW—a universally available, inex­ pensive parameter—as an independent predictor in the present study offers a more immediately translatable finding for resource-limited settings. This study has several limitations that merit acknowledgment. First, the retrospective, singlecenter design limits the generalizability of the find­ ings and introduces the potential for selection bias and incomplete data capture. Second, the sample size, while adequate for the primary analyses (n = 226), yielded a relatively small imaging subgroup (n = 92) with markedly unequal group sizes for the 10 neurological dysfunction comparison (84 vs. 8), which limits the statistical power to detect differences in imaging parameters and increases the risk of type II error. Third, the diagnosis of pure SAE relied on clinical judgment by the examining neurologist, and despite systematic exclusion criteria, some degree of subjectivity in the attribution of enceph­ alopathy etiology is inherent, particularly in patients with overlapping metabolic derangements. Fourth, more specific biomarkers of brain injury (S100β, NSE) and advanced neuroimaging modalities (MRI) were not available, which would have provided greater sensitivity in detecting subtle structural and functional brain abnormalities. Fifth, brain CT visual rating scales (GCA, Fazekas, MTA Schel­ tens) were assessed by a single investigator; interrater reliability was not evaluated, although these visual scales have generally shown acceptable interrater agreement in the literature. Finally, survival was assessed only at hospital discharge, and longerterm outcomes including cognitive recovery, func­ tional status, and post-discharge mortality were not captured. This study highlights the high burden of neuro­ logical dysfunction in emergency department sepsis patients and identifies simple, routinely available clinical, laboratory, and imaging markers associated with pure sepsis-associated encephalopathy. Prosp­ ective, multicenter validation studies are warranted to confirm these associations, develop composite bedside prediction tools, and ultimately determine whether early identification of SAE-prone patients can translate into targeted interventions that improve neurological outcomes and survival. Acknowledgments Grammarly (Grammarly Inc., San Francisco, CA, USA) was used to assist with English language editing. All suggested revisions were individually reviewed, verified, and approved by the authors, who take full responsibility for the final content and scientific accuracy of the manuscript. Funding No funding was received. Data availability The datasets used and/or analysed during the current study are available from the corresponding Juntendo Medical Journal, 2026 author on reasonable request. Author contributions FS and ES conceptualized the study, collected and curated the clinical data, performed the statis­ tical analysis, and drafted the manuscript. SF performed the neurological examinations, estab­ lished the diagnoses of neurological dysfunction and pure sepsis-associated encephalopathy, retro­ spectively reviewed all brain CT scans, and contrib­ uted to data interpretation. DB supervised the study design, contributed to data interpretation and critical revision of the manuscript, and provided final approval of the submitted version. All authors read, revised, and approved the final version of the manuscript and agree to be accountable for all aspects of the work. Conflicts of interest statement The authors declare that they have not conflict of interest. Author Scarlatescu Ecaterina, one of the Editorial Board members of JMJ was not involved in the peer review or decision-making process for this paper. References 1) Gofton TE, Young GB: Sepsis-associated encephalop­ athy. Nat Rev Neurol, 2012; 8: 557-566. 2) Mazeraud A, Righy C, Bouchereau E, Benghanem S, Bozza FA, Sharshar T: Septic-Associated Encephalop­ athy: a Comprehensive Review. Neurotherapeutics, 2020; 17: 392-403. 3) Zampieri FG, Park M, Machado FS, Azevedo LC: Sepsisassociated encephalopathy: not just delirium. Clinics (Sao Paulo), 2011; 66: 1825-1831. 4) Chaudhry N, Duggal AK: Sepsis Associated Encepha­ lopathy. Adv Med, 2014; 2014: 762320. 5) Chen J, Shi X, Diao M, et al: A retrospective study of sepsis-associated encephalopathy: epidemiology, clin­ ical features and adverse outcomes. BMC Neurol, 2020; 20: 77. 6) Meng L, Liu J, Wang P, et al: Risk factors of sepsis-as­ sociated encephalopathy in sepsis patients with pulmo­ nary infection: from the secondary analysis of a clus­ ter-randomized controlled trial. Respir Med, 2026; 252: 108658. 7) Becker AE, Teixeira SR, Lunig NA, et al: SepsisRelated Brain MRI Abnormalities Are Associated With Mortality and Poor Neurological Outcome in Pediatric Sepsis. Pediatr Neurol, 2022; 128: 1-8. 8) Rombi C, Klinkmann G, Hmeydia G, et al: Sepsis is associated with radiological lesions in patients with primary brain injuries: The result of a bicentric retro­ spective cohort. Anaesth Crit Care Pain Med, 2026; 45: 101635. 9) Singer M, Deutschman CS, Seymour CW, et al: The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA, 2016; 315: 801-810. 10) Hobden G, Colbourne E, Pendlebury ST, Demeyere N: Reliability of the global cortical atrophy visual rating scale applied to computed tomography versus magnetic resonance imaging scans in acute stroke. Eur Stroke J, 2024; 45: 1549-1556. 11) Fazekas F, Chawluk JB, Alavi A, Hurtig HI, Zimmerman RA: MR signal abnormalities at 1.5 T in Alzheimer's dementia and normal aging. AJR Am J Roentgenol, 1987; 149: 351-356. 12) Ferguson KJ, Cvoro V, MacLullich AMJ, et al: Visual Rating Scales of White Matter Hyperintensities and Atrophy: Comparison of Computed Tomography and Magnetic Resonance Imaging. J Stroke Cerebrovasc Dis, 2018; 27: 1815-1821. 13) van Westen D, Stomrud E, Lorenzini L, et al: Medial temporal lobe atrophy is associated with age and pathologies, especially small vessel disease. Sci Rep, 2026; 16: 3912. 14) Claus JJ, Staekenborg SS, Holl DC, et al: Practical use of visual medial temporal lobe atrophy cut-off scores in Alzheimer's disease: Validation in a large memory clinic population. Eur Radiol, 2017; 27: 3147-3155. 15) Fleuren KJR, Koehler PJ, Hoff EI: The History of the Evans Index. Eur Neurol, 2025; 88: 92-96. 16) Jaraj D, Rabiei K, Marlow T, Jensen C, Skoog I, Wikkelsø C: Estimated ventricle size using Evans index: refer­ ence values from a population-based sample. Eur J Neurol, 2017; 24: 468-474. 17) Barr AN, Heinze WJ, Dobben GD, Valvassori GE, Sugar O: Bicaudate index in computerized tomography of Huntington disease and cerebral atrophy. Neurology, 1978; 28: 1196-1200. 18) Zhou T, Tian X, Wang W, Chu Z: Diagnostic models for sepsis-associated encephalopathy: a comprehensive systematic review and meta-analysis. Front Neurol, 2025; 16: 1645397. 19) Sprung CL, Peduzzi PN, Shatney CH, et al: Impact of encephalopathy on mortality in the sepsis syndrome. The Veterans Administration Systemic Sepsis Cooper­ ative Study Group. Crit Care Med, 1990; 18: 801-806. 20) Crippa IA, Taccone FS, Wittebole X, Martin-Loeches I: The Prognostic Value of Brain Dysfunction in Criti­ cally Ill Patients with and without Sepsis: A Post Hoc Analysis of the ICON Audit. J Clin Med, 2021; 11: 530. 21) Sonneville R, de Montmollin E, Poujade J, et al: Poten­ tially modifiable factors contributing to sepsis-associ­ ated encephalopathy. Intensive Care Med, 2017; 43: 1075-1084. 22) He TT, Jiao TQ, An XM: Risk prediction models for sepsis-associated encephalopathy: a systematic evalu­ ation and meta-analysis. PeerJ, 2026; 14: e20770. 23) Shen D, Zhou S, Wu H, Zhang W, Li Z, Sun J: Associa­ tions between Glasgow Coma Scale trajectories and 28-day survival rate in patients with sepsis-associated encephalopathy: insights from longitudinal group trajec­ tory modeling. Front Neurol, 2025; 16: 1607946. 24) Zhang L, Yu C-H, Guo K-P, Huang C-Z, Mo L-Y: Prognostic role of red blood cell distribution width in patients with sepsis: a systematic review and metaanalysis. BMC Immunol, 2020; 21: 40. 25) Singh S, Diwakar N, Jha R, Vatsa P, Ahmad A: Assessing the Prognostic Accuracy of Red Cell Distribution Width for Predicting Mortality in Sepsis Patients: A Compar­ 11 Scarlatescu F, et al: Early correlates of sepsis-associated encephalopathy in the emergency department ative Study With Serum Lactate and the Quick Sequen­ tial Organ Failure Assessment (qSOFA) Score. Cureus, 2025; 17: e80056. 26) Bateman RM, Sharpe MD, Singer M, Ellis CG: The Effect of Sepsis on the Erythrocyte. Int J Mol Sci, 2017; 18: 1932. 27) De Backer D, Verdant C, Chierego M, Koch M, Gullo A, Vincent JL: Effects of drotrecogin alfa activated on microcirculatory alterations in patients with severe sepsis. Crit Care Med, 2006; 34: 1918-1924. 28) Ince C: The microcirculation is the motor of sepsis. Crit Care, 2005; 9 Suppl 4: S13-S19. 29) Joosse HJ, van Oirschot BA, Kooijmans SAA, et al: In-vitro and in-silico evidence for oxidative stress as drivers for RDW. Sci Rep, 2023; 13: 9223. 30) Liu H, Zhang T, Zhang L, Zhong Y: Neuroinflamma­ tory Mechanisms of Adult Sepsis-Associated Enceph­ alopathy: Implications for Blood-Brain Barrier Disrup­ tion and Oxidative Stress. Diagnostics (Basel), 2025; 15: 873. 31) Nian W, Tao W, Zhang H: Review of research progress in sepsis-associated acute kidney injury. Front Mol Biosci, 2025; 12: 1603392. 32) Pendlebury ST, Thomson RJ, Welch SJV, Kuker W, 12 Rothwell PM, Oxford Vascular Study: Utility of white matter disease and atrophy on routinely acquired brain imaging for prediction of long-term delirium risk: population-based cohort study. Age Ageing, 2021; 51: afab200. 33) Li Y, Chen JQ, Wang H, Yang JJ, Ji MH: Frontal, temporal, cerebellar changes link to sepsis survivors' cognitive issues: A resting state functional magnetic resonance imaging study. World J Psychiatry, 2025; 15: 108861. 34) Nakae R, Sekine T, Tagami T, et al: Rapidly progres­ sive brain atrophy in septic ICU patients: a retrospec­ tive descriptive study using semiautomatic CT volum­ etry. Crit Care, 2021; 25: 411. 35) Gu W, Zhong J, Lyu C, et al: An approach for the emer­ gency diagnosis and treatment of sepsis-associated encephalopathy in elderly individuals: a literature review. World J Emerg Med, 2025; 16: 415-422. 36) Tian E, Guo Y, Chen Y, Li W, Zhang X, Wang G: Predictive model of the occurrence of sepsis-associated encephalopathy in sepsis patients based on the combi­ nation of IFN-γ, TNF-α and CD4+/CD8+ ratio. Medi­ cine (Baltimore), 2025; 104: e45287.