Dynamic Mean Arterial Pressure Trajectories and Risk of Delirium in Patients with Sepsis: A Retrospective Cohort Study

Dynamic Mean Arterial Pressure Trajectories and Risk of Delirium in Patients with Sepsis: A Retrospective Cohort Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Dynamic Mean Arterial Pressure Trajectories and Risk of Delirium in Patients with Sepsis: A Retrospective Cohort Study wei song, yijiong zhang, jian wan, ye yuan, si peng, conghui fan, and 8 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8759359/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Background Delirium is a serious complication in septic ICU patients, often related to cerebral hypoperfusion and hemodynamic instability. However, the effect of dynamic blood pressure patterns on delirium risk remains unclear. Therefore, we sought to investigate the association between mean arterial pressure (MAP) trajectories during early ICU stay and the subsequent development of delirium in patients with sepsis. Methods We retrospectively analyzed 1,055 adults with sepsis from the Medical Information Mart for Intensive Care IV(MIMIC-IV)database. MAP during the first 24 hours after ICU admission was recorded every 2 hours. Group-based trajectory modeling identified distinct MAP patterns. The primary outcome was delirium during ICU stay. Associations between MAP trajectories and delirium risk were evaluated using multivariable Fine–Gray and Cox models, considering 7-day mortality as a competing event. Results Seven distinct MAP trajectories were identified, showing substantial interindividual variability in early hemodynamic patterns. Classes 4 and 6, characterized by rising and sharply declining MAP, exhibited the highest delirium risk, with cumulative incidences of about 75% and 65% by 120 hours. Using Class 5 (persistently low MAP) as the reference, multivariable analyses showed hazard ratios of 2.456–2.856 for Class 4 and 2.114–2.682 for Class 6 (all p < 0.05). Classes 1 and 5 had the lowest risk, while Classes 2, 3, and 7 showed intermediate risk. Subgroup analyses confirmed consistent associations across demographics and interventions. Conclusions Early MAP trajectories are strongly associated with delirium risk in sepsis. Acute hemodynamic deterioration, especially rising or sharply declining MAP, identifies high-risk patients. sepsis delirium MAP trajectory analysis ICU Figures Figure 1 Figure 2 Figure 3 Introduction Sepsis is a dysregulated host response to infection that leads to life-threatening organ dysfunction and remains a major cause of morbidity and mortality in ICUs worldwide. [ 1 – 2 ] Despite advances in critical care, sepsis continues to impose a substantial clinical and economic burden. [ 1 ] Early recognition and hemodynamic optimization are essential for improving patient outcomes. Delirium is the most common form of acute brain dysfunction in critically ill patients and is characterized by disturbances in consciousness, attention, orientation, and perception, including hallucinations and delusions. [3] Its development is multifactorial and may be triggered by acute illness, medication effects or withdrawal, trauma, or surgery. [4] Additional contributing factors include pain, sleep disruption, surgical stress, anesthesia, concomitant medications, inflammatory responses to tissue injury, and the release of inflammatory mediators secondary to cerebral hypoperfusion. [5] Delirium in ICU patients has been shown to significantly increase healthcare utilization, with medical costs rising by at least 20% compared with patients without delirium. [6] Moreover, sepsis-associated delirium is strongly linked to long-term cognitive impairment after hospital discharge, with affected patients experiencing accelerated cognitive decline and markedly reduced quality of life. [3,7] Given its high incidence and profound impact on patient outcomes, identifying modifiable risk factors is essential for effective prevention and management. Early recognition and targeted intervention remain critical to improving the prognosis of septic patients at risk of delirium. The Surviving Sepsis Campaign (SSC) guidelines recommend maintaining a MAP of at least 65 mmHg during initial resuscitation. [8] However, interindividual variability in cerebral autoregulation means that a uniform MAP target may not ensure adequate cerebral perfusion for all patients. Evidence from perioperative and critical care studies suggests that both hypotension and excessive blood pressure fluctuations increase the risk of delirium. [7,9–11] Nevertheless, most previous studies have focused on static or averaged MAP values, which fail to capture the dynamic nature of blood pressure changes in the ICU. group-based trajectory modeling (GBTM) offers a novel approach to characterize distinct temporal patterns of physiological variables. Applying trajectory analysis to MAP dynamics may help identify subgroups of septic patients at high risk of delirium and improve understanding of the relationship between hemodynamic instability and acute brain dysfunction. Therefore, this study aimed to investigate the association between early MAP trajectories and the development of delirium in critically ill patients with sepsis. Materials and methods Study design This retrospective cohort study utilized data from the MIMIC-IV(version 3.1), a large, publicly accessible database comprising comprehensive clinical records of patients admitted to the intensive care units of Beth Israel Deaconess Medical Center between 2008 and 2019. 1 Database access was granted following completion of the required data-use training, and all data were fully de-identified. Ethical approval was therefore waived by the institutional review boards of both the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. The study was conducted in accordance with the Declaration of Helsinki and is reported in compliance with the STROBE guidelines for observational studies [12]. Study population Patients were identified according to the Sepsis-3 definition, in which sepsis is characterized by suspected or confirmed infection in combination with an increase in the Sequential Organ Failure Assessment (SOFA) score of at least 2 points. [8] Only adult patients aged 18 years or older were included. In cases of repeated hospital admissions, only the first ICU admission was considered for analysis. Additional eligibility criteria required an ICU stay exceeding 7 days and adequate hemodynamic monitoring, defined as continuous arterial blood pressure recording for a minimum of 24 hours with no fewer than four valid measurements during this period. Patients were excluded if they had a documented history of stroke or intracranial hemorrhage, pre-existing cognitive impairment or dementia, prior use of psychotropic medications before ICU admission, or incomplete delirium assessment records. To characterize early hemodynamic dynamics, GBTM was applied to identify distinct patterns of MAP changes during the first 24 hours following ICU admission. MAP values were obtained from invasive arterial pressure monitoring and calculated using the standard formula: one-third systolic blood pressure plus two-thirds diastolic blood pressure. Multiple trajectory models specifying between one and eight latent groups were constructed (Table 1). Model selection was guided by goodness-of-fit indices, including the Akaike information criterion and Bayesian information criterion, together with clinical interpretability. Each patient was assigned to the trajectory group corresponding to the highest posterior probability, with an average posterior probability greater than 0.7 considered indicative of satisfactory classification accuracy. Kernel density plots were generated to illustrate the distribution of MAP values across the identified trajectory groups. Outcomes The primary outcome was the occurrence of delirium during the ICU stay. Delirium was assessed using the CAM-ICU and documented in the electronic medical records. Patients were classified as having delirium if they had at least one positive CAM-ICU assessment during their ICU stay. Baseline information collection Baseline characteristics encompassed demographic information (age, sex, height, and weight); disease severity evaluated using the Acute Physiology and Chronic Health Evaluation II (APACHE II), SOFA, and Glasgow Coma Scale (GCS); and laboratory variables, including white blood cell count (WBC), hemoglobin, platelet count, coagulation indices such as prothrombin time (PT), activated partial thromboplastin time (APTT), and international normalized ratio (INR), triglycerides, blood glucose, albumin, serum electrolytes (potassium and sodium), arterial blood gas parameters including pH, arterial partial pressure of oxygen (PaO₂), arterial partial pressure of carbon dioxide (PaCO₂), and lactate. Baseline comorbidities comprised hypertension, diabetes, acute kidney injury (AKI), pneumonia, heart failure, and myocardial infarction. In addition, ICU-related interventions, including the administration of vasoactive agents, sedatives, and corticosteroids, were recorded. Study outcomes included the occurrence of delirium during ICU stay and 28-day mortality. All baseline variables were collected within the first 24 hours after ICU admission. Statistical analysis Missing values in covariates were imputed using the K-nearest neighbors (KNN) algorithm. Variables with more than 20% missing data were excluded from the analysis. MAP measurements collected within the first 24 hours after ICU admission were included and aggregated into 2-hour intervals, with at least four valid measurements required per patient. GBTM was used to identify distinct MAP trajectory patterns over the 24-hour period. Patients were assigned to trajectory groups based on the highest posterior probability. Model performance was evaluated using the Akaike information criterion, Bayesian information criterion, sample-size adjusted BIC, and entropy. After comparison of models with two to eight trajectories, a seven-trajectory solution was selected based on overall model fit and clinical interpretability. All trajectory analyses were conducted using the lcmm package in R. Continuous variables were assessed for normality using the Shapiro–Wilk test and summarized as mean ± SD or median (IQR), as appropriate. Group comparisons were performed using Student’s t test or ANOVA for normally distributed data, and the Mann–Whitney U or Kruskal–Wallis test for non-normally distributed data. Categorical variables were compared using the chi-square test. The trajectory group with the lowest incidence of delirium served as the reference group. Kaplan–Meier curves were constructed to estimate cumulative incidence. Given the presence of competing risks due to early death, the Fine–Gray subdistribution hazard model was applied, with between-group differences assessed using the log-rank test. Four multivariable models were sequentially constructed to examine the association between MAP trajectories and delirium. Model 1 included trajectory groups only; Model 2 adjusted for laboratory variables; Model 3 further adjusted for SOFA and GCS scores; and Model 4 additionally accounted for major comorbidities. Covariates were selected based on clinical relevance and multivariable regression results. Statistical analyses were performed using R software (version 4.5.1), with a two-sided p value < 0.05 considered statistically significant. Results Study cohort From the MIMIC IV database, 31,910 patients with sepsis were initially identified. After screening, 7,497 patients met the inclusion criteria. Patients who did not meet the study requirements or had missing delirium assessment data were excluded. Finally, 1,055 patients were included in the analysis and categorized into seven trajectory groups based on their blood pressure changes during the first 24 hours after ICU admission. These groups represented distinct temporal patterns of blood pressure variation (Fig. 1 ). MAP trajectory subphenotypes in sepsis patients Taking statistical and clinical interpretability into consideration, our model eventually identified seven unique trajectory groups with relatively low AIC, BIC, and SABIC values, as well as relatively high log-likelihood ratios (Table 1). The mean posterior probabilities of group membership for the group members were all above 70%, further supporting a great overall fit of the 7-group model (Table 2 ). The fixed effects of the seven-class longitudinal model are detailed in Table 3 . Latent class growth analysis(LCGA)identified seven distinct dynamic trajectories of MAP within the first 24 hours after ICU admission (Fig. 2 ). Class 1 (n = 202, 19.1%) showed moderate MAP values, increasing slightly from approximately 83 mmHg to 86 mmHg. Class 2 (n = 112, 10.6%) started with high MAP around 98 mmHg and gradually declined to 88 mmHg. Class 3 (n = 59, 5.6%) remained at a consistently elevated level, decreasing modestly from 109 mmHg to 106 mmHg. Class 4 (n = 23, 2.2%) exhibited a rising pattern, increasing from 82 mmHg to 108 mmHg. Class 5 (n = 174, 16.5%) demonstrated persistently low MAP, remaining stable around 63mmHg. Class 6 (n = 56, 5.3%) started from 98 mmHg and showed a sharp decline to approximately 65 mmHg. Class 7 (n = 429, 40.7%) represented the largest subgroup, with MAP decreasing slightly from 76 mmHg to 72 mmHg. These heterogeneous trajectories indicate substantial variability in early hemodynamic patterns among patients with sepsis. Comparisons of patient characteristics between trajectory groups Baseline characteristics differed significantly across the seven MAP trajectory classes (Table 4). Patients in Class 4 were the youngest (median age 58 years), whereas those in Class 5 and Class 6 were the oldest (median age 70.5 and 71.0 years, respectively). Disease severity varied across classes, with the highest APACHE II and SOFA scores observed in Class 5–7 and the lowest in Class 2–4. Laboratory tests indicated that patients in Class 5–7 were more likely to have lower albumin and sodium levels, as well as abnormal coagulation parameters. Inflammatory markers also differed, with Class 7 exhibiting the highest white blood cell count. Arterial blood gas analysis showed variations in pH and lactate levels among groups. Regarding comorbidities, the prevalence of hypertension, acute kidney injury, heart failure, and diabetes differed across trajectory classes. The incidence of delirium at the first assessment also varied significantly: Class 6 had the highest incidence (37.5%), followed by Class 4 (34.8%) and Class 7 (27.5%), whereas Class 5 had the lowest incidence (18.4%), with Class 1 (20.8%) and Class 3 (20.3%) showing relatively lower rates. Patients in Class 5–7 were more likely to receive vasoactive agents and neuromuscular blockers. No significant differences were observed in gender distribution, pneumonia, myocardial infarction, or ICU 28-day mortality. Univariate and multivariate analysis Figure 3 A presents the Kaplan-Meier curves demonstrating the cumulative incidence of delirium among the seven hemodynamic trajectory classes. A significant divergence in delirium risk was observed across the groups (log-rank test, p = 0.039). Class 4 and Class 6 exhibited high-risk profiles during the early phase, with a sharp increase in delirium incidence within the first 24 hours after ICU admission, reaching approximately 75% and 60%, respectively—significantly higher than other classes. In contrast, Class 1 and Class 5 were associated with considerably lower risk, showing gradually rising curves with cumulative incidence rates of 40% and 45% at 120 hours. Classes 2, 3, and 7 demonstrated intermediate risk patterns. Notably, Class 7, despite being the largest subgroup, displayed a persistently increasing risk throughout the mid-to-late observation period. The number-at-risk table confirms sufficient sample sizes during the critical initial phase, supporting the reliability of these findings. These results underscore that early hemodynamic instability, particularly the acute deterioration patterns represented by Class 4 and Class 6, serves as a critical predictor of delirium development in septic patients. Figure 3 B presents the cumulative incidence function (CIF) curves with 7-day mortality as a competing risk. Gray’s test indicated statistically significant differences in delirium risk among the seven hemodynamic trajectory classes (p = 0.039). Class 4 and Class 6 exhibited high-risk profiles, with a rapid increase in delirium probability during the initial 24 hours. The cumulative incidence of delirium in Class 4 reached approximately 75% by 120 hours, while Class 6 reached approximately 65%. In contrast, Class 1 and Class 5 demonstrated lower and more gradual increases in delirium probability, with 168-hour cumulative incidence rates of approximately 25% and 35%, respectively. Cox proportional hazards regression analysis demonstrated significant differences in delirium risk across MAP trajectory classes, as shown in Table 5 . Notably, Class 4 and Class 6 were consistently associated with markedly increased delirium risk in all models. Using Class 5 as the reference, the hazard ratios for Class 4 ranged from 2.456 to 2.856 (p < 0.05), and for Class 6 from 2.114 to 2.682 (p < 0.01), indicating that patients in these two trajectories were at the highest and most robust risk. In contrast, Classes 2, 3, and 7 showed moderate risk elevation in some models, whereas Class 1 was not significantly associated with delirium. Subgroup analysis Subgroup analyses were conducted to examine the robustness of the association between MAP trajectory classes and delirium across different clinical characteristics (Table 6 ). No significant interaction effects were observed among subgroups stratified by age, gender, use of sedatives, vasoactive agents, hypertension, or diabetes (all P for interaction > 0.05). Discussion In this study, we analyzed the trajectories of MAP during the first 24 hours after ICU admission in patients with sepsis using data from the MIMIC-IV database. Seven distinct MAP trajectory patterns were identified. The results demonstrated that both the rising (Class 4) and sharply declining (Class 6) MAP trajectories were significantly associated with the occurrence of delirium. This association remained robust after multivariable adjustment and was further confirmed in multiple subgroup analyses. In contrast, patients with persistently low MAP (Class 5) exhibited the lowest incidence of delirium, while those with moderately stable MAP (Class 1) also showed a relatively low risk. The remaining trajectory types were associated with intermediate risk levels. These findings suggest that the development of delirium in sepsis is not solely related to the absolute level of MAP but is also closely linked to the dynamic fluctuations of blood pressure over time. Cerebral autoregulation (CA) is a key mechanism that maintains stable cerebral perfusion, allowing cerebral blood flow to remain relatively constant despite fluctuations in systemic blood pressure. 2 However, under critical conditions—particularly in patients with sepsis, traumatic brain injury, or stroke—CA function is often impaired, rendering cerebral perfusion pressure passively dependent on mean arterial pressure (MAP). Under such circumstances, large fluctuations in MAP can directly affect cerebral blood flow and oxygen delivery, exacerbating neuronal injury. [13–16] Previous studies have also shown that increased blood pressure variability (BPV) is a stronger predictor of postoperative delirium than low blood pressure, suggesting that dynamic instability independently contributes to cerebral dysfunction. [17] Excessive blood pressure fluctuations not only reflect systemic circulatory instability but may also directly damage the vasculature through hemodynamic mechanisms. [15–18] Studies have shown that prolonged or pronounced blood pressure variability can promote atherosclerosis and increase arterial stiffness, thereby impairing cerebral microvascular perfusion and structural integrity. [18] Chronic high variability in blood pressure may induce cumulative damage throughout the arterial tree down to smaller vessels, further compromising blood flow to arterioles and capillaries, particularly in regions susceptible to ischemic-hypoxic injury, including subcortical white matter and the cerebral cortex. [19] Endothelial cells are highly sensitive to shear stress induced by blood flow; [20] abrupt changes in shear stress can trigger endothelial apoptosis and senescence via PKC ζ, JNK-MAPK, p53, and unfolded protein response pathways. [18,20–21] In contrast, stable blood flow exerts protective effects through nitric oxide synthase and antioxidant enzyme pathways. [18,22] Burkhart et al. further highlighted that cerebral perfusion in sepsis-associated encephalopathy (SAE) is regulated by both macrocirculatory and microcirculatory hemodynamics, with blood–brain barrier (BBB) disruption being a key mechanism. [23] Transcranial Doppler studies by Pfister et al. also confirmed that CA is impaired in SAE patients, rendering cerebral perfusion more susceptible to MAP fluctuations. [24] Collectively, these findings suggest that dynamic blood pressure instability may contribute to cerebral hypoperfusion and delirium through endothelial dysfunction and BBB disruption. In the present study, Class 4 and Class 6 MAP trajectories, representing rapidly rising and sharply declining patterns, likely reflect this impaired cerebral blood flow regulation, which may explain their strong association with increased delirium risk. The optimal mean arterial pressure (MAP) target in patients with septic shock remains controversial. Current guidelines recommend maintaining MAP at approximately 65 mmHg; [8] however, the SEPSISPAM trial demonstrated no significant difference in overall mortality between high MAP (80–85 mmHg) and low MAP (65–70 mmHg) groups, with the exception that patients with a history of hypertension in the high-target group required less renal replacement therapy. [25] Notably, Deruddre et al. reported that organ perfusion responses vary considerably among individuals once MAP exceeds 65 mmHg. This inter-individual variability is particularly relevant for cerebral perfusion and neurological function. In patients with impaired brain function, a uniform MAP target may result in either excessive or insufficient cerebral perfusion, potentially triggering ischemic or congestive neuronal injury and increasing the risk of delirium. [7] In our study, we observed that patients whose MAP remained stable 63 mmHg exhibited the lowest risk of delirium, while 28-day mortality did not differ significantly compared with other MAP ranges. These findings suggest that fixed MAP targets may overlook individual differences in cerebral autoregulation, thereby amplifying the detrimental effects of blood pressure fluctuations on brain function. Our findings suggest that the prediction and prevention of delirium should shift from reliance on static blood pressure measurements to dynamic monitoring. Traditional management strategies targeting a single MAP value may overlook blood pressure variability and inter-individual differences in cerebral autoregulation, potentially failing to mitigate the risk of neuronal injury. Blood pressure trajectory analysis provides continuous, dynamic hemodynamic information, facilitating early identification of high-risk patients; among these, Class 4 (rising) and Class 6 (sharply declining) trajectories should be prioritized for close monitoring. In hemodynamic management, attention should be given to minimizing blood pressure fluctuations through gradual titration of vasoactive agents, cautious fluid resuscitation, and optimization of sedation strategies, thereby avoiding abrupt changes that may compromise cerebral perfusion. Individualized risk assessment—taking into account pre-existing hypertension, cerebral autoregulation status, and organ perfusion—should guide the determination of patient-specific MAP targets. Future blood pressure management strategies for septic patients should be more individualized, incorporating CA assessment and near-infrared spectroscopy (NIRS) monitoring to dynamically identify each patient’s optimal MAP range. This approach aims to preserve cerebral perfusion while reducing the risk of delirium. Furthermore, integrating blood pressure trajectories with other clinical parameters—such as oxygen saturation, lactate levels, and neurological monitoring indices—may facilitate a multidimensional, real-time hemodynamic management strategy, providing more precise neuroprotective interventions for patients with septic shock. Strengths and Limitations The strengths of this study lie in the use of a large-scale cohort of sepsis patients and the application of advanced blood pressure trajectory modeling, which allowed for a comprehensive characterization of early dynamic blood pressure changes. Furthermore, the robustness of the findings was confirmed through multivariable and competing-risk analyses. Nevertheless, several limitations should be acknowledged. First, the retrospective design limits the ability to fully account for potential confounding factors. Second, delirium identification relied on electronic health record documentation, which may be subject to underdiagnosis or misclassification. Third, direct measurements of cerebral perfusion and cerebral autoregulation were not available, and the inferred physiological mechanisms require validation in prospective studies. Additionally, prior studies have suggested that blood pressure variability may be influenced by seasonal and environmental factors, which warrant further investigation in future research. Conclusion In conclusion, our study demonstrates that early dynamic MAP trajectories are closely linked to the risk of delirium in patients with sepsis. Specifically, acute hemodynamic deterioration—manifested as rapidly rising or sharply declining MAP patterns—can identify individuals at high risk. These findings underscore the clinical importance of continuous, individualized blood pressure monitoring and timely hemodynamic interventions to prevent delirium in critically ill septic patients. Future prospective studies are warranted to validate these observations and to explore targeted strategies for optimizing blood pressure stability in this vulnerable population. Abbreviations Abbreviation Full term MAP Mean arterial pressure ICU Intensive care unit MIMIC-IV Medical Information Mart for Intensive Care IV GBTM Group-based trajectory modeling LCGA Latent class growth analysis CAM-ICU Confusion Assessment Method for the Intensive Care Unit SOFA Sequential Organ Failure Assessment APACHE II Acute Physiology and Chronic Health Evaluation II GCS Glasgow Coma Scale WBC White blood cell count PT Prothrombin time APTT Activated partial thromboplastin time INR International normalized ratio AKI Acute kidney injury SBP Systolic blood pressure DBP Diastolic blood pressure PaO₂ Arterial partial pressure of oxygen PaCO₂ Arterial partial pressure of carbon dioxide BPV Blood pressure variability CA Cerebral autoregulation SAE Sepsis-associated encephalopathy BBB Blood–brain barrier NIRS Near-infrared spectroscopy KNN K-nearest neighbors AIC Akaike information criterion BIC Bayesian information criterion SABIC Sample-size adjusted Bayesian information criterion CIF Cumulative incidence function HR Hazard ratio CI Confidence interval IRB Institutional Review Board STROBE Strengthening the Reporting of Observational Studies in Epidemiology Declarations Ethical Statement This study utilized de-identified data from the publicly available MIMIC-IV database. Ethical approval and individual patient consent were not required, as confirmed by the Institutional Review Board of the Beth Israel Deaconess Medical Center. Access to the database was approved after completion of the CITI “Data or Specimens Only Research” course (Record ID: 68950718). Clinical trial number Not applicable Human Ethics and Consent to Participate declarations Not applicable Competing interest Competing interest All the authors have disclosed that they do not have any conflicts of interest Availability of data and material Data supporting the findings of this study are available from the corresponding author upon reasonable request, and all other relevant data are included in the main text. Funding Statement This study was supported by the following programs: Outstanding Young Medical Talent Program of the Pudong New Area Health Commission, Shanghai, China (No. PWRq2025-26) Discipline Leader Training Program of the Pudong New Area Health System, Shanghai (No. PWRd2024-12) The Project of Key Medical Discipline Group Construction in Shanghai Pudong New Area (No. PWZxq2022-13) The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author Contributions (CRediT roles) Wei Song (H.Z.): Conceptualization; Study design; Data acquisition, analysis, and interpretation; Writing – original draft; Visualization; Project administration; Final approval; Accountability. (First author) Yijiong Zhang , Jian Wan ,Ye Yuan , Si Peng: Statistical modeling; Figure preparation and editing; Results interpretation; Writing – review and editing; Funding acquisition. (These authors contributed equally with the first author) Conghui Fan, Song Chen, Qian Zhang, Fei Zhong: Assistance with data analysis; Guidance on manuscript revision. Tao Zhang: Software implementation; Data processing; Visualization. Xueke Wu: Technical implementation; Data verification. Zhen Han: Clinical expertise; Quality assurance. Qingzhong Zhao: Manuscript critique; Final approval; Accountability. Haijun Zhu: Critically revised the manuscript, and takes responsibility for the integrity of the work. Acknowledgment The authors would like to express their sincere gratitude to Weijian Zhang for his valuable software and technical support throughout the course of this study. We are also grateful to Prof. Jian Wan , Director of our department, for his insightful guidance and constructive suggestions on the research design and implementation. In addition, we would like to acknowledge Prof. Song Chen for his generous financial support, which made this work possible. References Song W, et al. Comorbidity of PaO(2)/FiO(2) and nonthyroidal illness syndrome synergistically predicts 28-day mortality in sepsis: A retrospective cohort study from the MIMIC-IV database. Sci Prog. 2025;108:368504251359356. Armstead WM. Cerebral Blood Flow Autoregulation and Dysautoregulation. Anesthesiol Clin. 2016;34:465–77. Tables Tables 1 to 6 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Tables.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 19 Apr, 2026 Reviewers agreed at journal 09 Apr, 2026 Reviewers invited by journal 12 Feb, 2026 Editor assigned by journal 12 Feb, 2026 Editor invited by journal 09 Feb, 2026 Submission checks completed at journal 08 Feb, 2026 First submitted to journal 08 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8759359","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":591656770,"identity":"eac5da75-b095-4ac0-b68e-f71a8cc6412d","order_by":0,"name":"wei song","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"wei","middleName":"","lastName":"song","suffix":""},{"id":591656771,"identity":"5a39d24f-ce11-4c76-b1ad-9850741bf70e","order_by":1,"name":"yijiong zhang","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"yijiong","middleName":"","lastName":"zhang","suffix":""},{"id":591656772,"identity":"d92bc5c4-7ac0-4332-b94e-3bb3c738f3e8","order_by":2,"name":"jian wan","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"jian","middleName":"","lastName":"wan","suffix":""},{"id":591656773,"identity":"957e1b99-22fb-41df-9be6-a1415e5bfd44","order_by":3,"name":"ye yuan","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"ye","middleName":"","lastName":"yuan","suffix":""},{"id":591656774,"identity":"46e964ad-5a78-4d0a-8a2c-9d391e912b88","order_by":4,"name":"si peng","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"si","middleName":"","lastName":"peng","suffix":""},{"id":591656775,"identity":"aa51dbf1-b39b-4a99-a1bb-0ba44ec6c43b","order_by":5,"name":"conghui fan","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"conghui","middleName":"","lastName":"fan","suffix":""},{"id":591656776,"identity":"3c0cf191-0d71-40fe-a8ca-0d4c68dda4de","order_by":6,"name":"song chen","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"song","middleName":"","lastName":"chen","suffix":""},{"id":591656777,"identity":"8a19e16d-2ad2-419a-a72b-b2b36ab3cdc3","order_by":7,"name":"qian zhang","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"qian","middleName":"","lastName":"zhang","suffix":""},{"id":591656778,"identity":"37c7f569-9dad-46d9-befa-71fda13f5a6a","order_by":8,"name":"fei zhong","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"fei","middleName":"","lastName":"zhong","suffix":""},{"id":591656779,"identity":"0b01cf1b-1aa6-4059-9483-752029b24807","order_by":9,"name":"tao zhang","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"tao","middleName":"","lastName":"zhang","suffix":""},{"id":591656780,"identity":"648f84f0-9db6-42dd-a7f7-58aa42f4d725","order_by":10,"name":"xueke wu","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"xueke","middleName":"","lastName":"wu","suffix":""},{"id":591656781,"identity":"dda02955-b623-4408-bdae-68c9048b650e","order_by":11,"name":"zhen han","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"zhen","middleName":"","lastName":"han","suffix":""},{"id":591656782,"identity":"cad71f0f-1a58-4f12-b9bc-bf4124346002","order_by":12,"name":"qingzhong zhao","email":"","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":false,"prefix":"","firstName":"qingzhong","middleName":"","lastName":"zhao","suffix":""},{"id":591656783,"identity":"08e7f564-5fd7-4dc9-ae5d-67b5bf9222bf","order_by":13,"name":"haijun zhu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1ElEQVRIiWNgGAWjYDACCQY2BgYDIGZvABJsDAkkaOE5QJIWMCOBSC3ys3vMHvwo4EvcLvn82eOCsjt5DOyHj27Ap4Vxzhlzwx4DtsSds3PMjWece1bMwJOWdgOfFmaJHDMJHqCWDbdz2KR52w4nNkjwmOHVwgbUIvkHpOXm8WfEaeEBapEG23KDwYw4LRISaWXSMgZsxhvOAPUC/ZLYRsgv8jOSt0m++XNMdsNxoMOAIZbYz374GF4tUHAMTDIzMByARRNBUIPQMgpGwSgYBaMAHQAAdWRJDXQuX6gAAAAASUVORK5CYII=","orcid":"","institution":"Shanghai Pudong New Area People's Hospital","correspondingAuthor":true,"prefix":"","firstName":"haijun","middleName":"","lastName":"zhu","suffix":""}],"badges":[],"createdAt":"2026-02-02 01:23:05","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8759359/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8759359/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102855761,"identity":"6776f85e-192e-4a4d-9040-1a2435e7eccc","added_by":"auto","created_at":"2026-02-17 14:57:29","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":125744,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of patient selection in the study.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8759359/v1/e119c9539cdbc9bd15fa6d7e.png"},{"id":102855727,"identity":"bc4374df-034c-4277-8f9a-9c66a718fb49","added_by":"auto","created_at":"2026-02-17 14:57:14","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":111804,"visible":true,"origin":"","legend":"\u003cp\u003eSeven classes identified by trajectories of MAP.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8759359/v1/9e7f4ae49f8af87b1c9900c8.png"},{"id":102855753,"identity":"461e484d-3dea-4793-9c6e-0323338da461","added_by":"auto","created_at":"2026-02-17 14:57:27","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":216104,"visible":true,"origin":"","legend":"\u003cp\u003eCumulative incidence curves by the Kaplan–Meier method. (A) Delirium incidence across seven MAP trajectory classes. (B) Delirium incidence across seven MAP trajectory classes adjusted for 7-day mortality as a competing risk.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8759359/v1/36422dfb36cbd6ed45b6634d.jpeg"},{"id":102855828,"identity":"73921cc8-228b-4739-9209-04bdbb72eaf4","added_by":"auto","created_at":"2026-02-17 14:57:43","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1154802,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8759359/v1/8f9a34b8-09f8-474d-8485-3b2208021a03.pdf"},{"id":102855741,"identity":"4814abd5-02f0-42f7-a415-ce6eb7342a89","added_by":"auto","created_at":"2026-02-17 14:57:16","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":41113,"visible":true,"origin":"","legend":"","description":"","filename":"Tables.docx","url":"https://assets-eu.researchsquare.com/files/rs-8759359/v1/e583bd6c48d3fb28e77a1c97.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Dynamic Mean Arterial Pressure Trajectories and Risk of Delirium in Patients with Sepsis: A Retrospective Cohort Study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eSepsis is a dysregulated host response to infection that leads to life-threatening organ dysfunction and remains a major cause of morbidity and mortality in ICUs worldwide. [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e] Despite advances in critical care, sepsis continues to impose a substantial clinical and economic burden. [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e] Early recognition and hemodynamic optimization are essential for improving patient outcomes.\u003c/p\u003e \u003cp\u003eDelirium is the most common form of acute brain dysfunction in critically ill patients and is characterized by disturbances in consciousness, attention, orientation, and perception, including hallucinations and delusions. [3] Its development is multifactorial and may be triggered by acute illness, medication effects or withdrawal, trauma, or surgery. [4] Additional contributing factors include pain, sleep disruption, surgical stress, anesthesia, concomitant medications, inflammatory responses to tissue injury, and the release of inflammatory mediators secondary to cerebral hypoperfusion. [5] Delirium in ICU patients has been shown to significantly increase healthcare utilization, with medical costs rising by at least 20% compared with patients without delirium. [6] Moreover, sepsis-associated delirium is strongly linked to long-term cognitive impairment after hospital discharge, with affected patients experiencing accelerated cognitive decline and markedly reduced quality of life. [3,7] Given its high incidence and profound impact on patient outcomes, identifying modifiable risk factors is essential for effective prevention and management. Early recognition and targeted intervention remain critical to improving the prognosis of septic patients at risk of delirium.\u003c/p\u003e \u003cp\u003e The Surviving Sepsis Campaign (SSC) guidelines recommend maintaining a MAP of at least 65 mmHg during initial resuscitation. [8] However, interindividual variability in cerebral autoregulation means that a uniform MAP target may not ensure adequate cerebral perfusion for all patients. Evidence from perioperative and critical care studies suggests that both hypotension and excessive blood pressure fluctuations increase the risk of delirium. [7,9\u0026ndash;11] Nevertheless, most previous studies have focused on static or averaged MAP values, which fail to capture the dynamic nature of blood pressure changes in the ICU.\u003c/p\u003e \u003cp\u003egroup-based trajectory modeling (GBTM) offers a novel approach to characterize distinct temporal patterns of physiological variables. Applying trajectory analysis to MAP dynamics may help identify subgroups of septic patients at high risk of delirium and improve understanding of the relationship between hemodynamic instability and acute brain dysfunction. Therefore, this study aimed to investigate the association between early MAP trajectories and the development of delirium in critically ill patients with sepsis.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy design\u003c/h2\u003e \u003cp\u003eThis retrospective cohort study utilized data from the MIMIC-IV(version 3.1), a large, publicly accessible database comprising comprehensive clinical records of patients admitted to the intensive care units of Beth Israel Deaconess Medical Center between 2008 and 2019.\u003csup\u003e1\u003c/sup\u003e Database access was granted following completion of the required data-use training, and all data were fully de-identified. Ethical approval was therefore waived by the institutional review boards of both the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center. The study was conducted in accordance with the Declaration of Helsinki and is reported in compliance with the STROBE guidelines for observational studies [12].\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStudy population\u003c/h3\u003e\n\u003cp\u003ePatients were identified according to the Sepsis-3 definition, in which sepsis is characterized by suspected or confirmed infection in combination with an increase in the Sequential Organ Failure Assessment (SOFA) score of at least 2 points. [8] Only adult patients aged 18 years or older were included. In cases of repeated hospital admissions, only the first ICU admission was considered for analysis. Additional eligibility criteria required an ICU stay exceeding 7 days and adequate hemodynamic monitoring, defined as continuous arterial blood pressure recording for a minimum of 24 hours with no fewer than four valid measurements during this period. Patients were excluded if they had a documented history of stroke or intracranial hemorrhage, pre-existing cognitive impairment or dementia, prior use of psychotropic medications before ICU admission, or incomplete delirium assessment records.\u003c/p\u003e \u003cp\u003eTo characterize early hemodynamic dynamics, GBTM was applied to identify distinct patterns of MAP changes during the first 24 hours following ICU admission. MAP values were obtained from invasive arterial pressure monitoring and calculated using the standard formula: one-third systolic blood pressure plus two-thirds diastolic blood pressure. Multiple trajectory models specifying between one and eight latent groups were constructed (Table\u0026nbsp;1). Model selection was guided by goodness-of-fit indices, including the Akaike information criterion and Bayesian information criterion, together with clinical interpretability. Each patient was assigned to the trajectory group corresponding to the highest posterior probability, with an average posterior probability greater than 0.7 considered indicative of satisfactory classification accuracy. Kernel density plots were generated to illustrate the distribution of MAP values across the identified trajectory groups.\u003c/p\u003e\n\u003ch3\u003eOutcomes\u003c/h3\u003e\n\u003cp\u003eThe primary outcome was the occurrence of delirium during the ICU stay. Delirium was assessed using the CAM-ICU and documented in the electronic medical records. Patients were classified as having delirium if they had at least one positive CAM-ICU assessment during their ICU stay.\u003c/p\u003e\n\u003ch3\u003eBaseline information collection\u003c/h3\u003e\n\u003cp\u003eBaseline characteristics encompassed demographic information (age, sex, height, and weight); disease severity evaluated using the Acute Physiology and Chronic Health Evaluation II (APACHE II), SOFA, and Glasgow Coma Scale (GCS); and laboratory variables, including white blood cell count (WBC), hemoglobin, platelet count, coagulation indices such as prothrombin time (PT), activated partial thromboplastin time (APTT), and international normalized ratio (INR), triglycerides, blood glucose, albumin, serum electrolytes (potassium and sodium), arterial blood gas parameters including pH, arterial partial pressure of oxygen (PaO₂), arterial partial pressure of carbon dioxide (PaCO₂), and lactate. Baseline comorbidities comprised hypertension, diabetes, acute kidney injury (AKI), pneumonia, heart failure, and myocardial infarction. In addition, ICU-related interventions, including the administration of vasoactive agents, sedatives, and corticosteroids, were recorded. Study outcomes included the occurrence of delirium during ICU stay and 28-day mortality. All baseline variables were collected within the first 24 hours after ICU admission.\u003c/p\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eMissing values in covariates were imputed using the K-nearest neighbors (KNN) algorithm. Variables with more than 20% missing data were excluded from the analysis. MAP measurements collected within the first 24 hours after ICU admission were included and aggregated into 2-hour intervals, with at least four valid measurements required per patient.\u003c/p\u003e \u003cp\u003eGBTM was used to identify distinct MAP trajectory patterns over the 24-hour period. Patients were assigned to trajectory groups based on the highest posterior probability. Model performance was evaluated using the Akaike information criterion, Bayesian information criterion, sample-size adjusted BIC, and entropy. After comparison of models with two to eight trajectories, a seven-trajectory solution was selected based on overall model fit and clinical interpretability. All trajectory analyses were conducted using the lcmm package in R.\u003c/p\u003e \u003cp\u003eContinuous variables were assessed for normality using the Shapiro\u0026ndash;Wilk test and summarized as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;SD or median (IQR), as appropriate. Group comparisons were performed using Student\u0026rsquo;s t test or ANOVA for normally distributed data, and the Mann\u0026ndash;Whitney U or Kruskal\u0026ndash;Wallis test for non-normally distributed data. Categorical variables were compared using the chi-square test.\u003c/p\u003e \u003cp\u003eThe trajectory group with the lowest incidence of delirium served as the reference group. Kaplan\u0026ndash;Meier curves were constructed to estimate cumulative incidence. Given the presence of competing risks due to early death, the Fine\u0026ndash;Gray subdistribution hazard model was applied, with between-group differences assessed using the log-rank test.\u003c/p\u003e \u003cp\u003eFour multivariable models were sequentially constructed to examine the association between MAP trajectories and delirium. Model 1 included trajectory groups only; Model 2 adjusted for laboratory variables; Model 3 further adjusted for SOFA and GCS scores; and Model 4 additionally accounted for major comorbidities. Covariates were selected based on clinical relevance and multivariable regression results. Statistical analyses were performed using R software (version 4.5.1), with a two-sided p value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 considered statistically significant.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n \u003ch2\u003eStudy cohort\u003c/h2\u003e\n \u003cp\u003eFrom the MIMIC IV database, 31,910 patients with sepsis were initially identified. After screening, 7,497 patients met the inclusion criteria. Patients who did not meet the study requirements or had missing delirium assessment data were excluded. Finally, 1,055 patients were included in the analysis and categorized into seven trajectory groups based on their blood pressure changes during the first 24 hours after ICU admission. These groups represented distinct temporal patterns of blood pressure variation (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eMAP trajectory subphenotypes in sepsis patients\u003c/h3\u003e\n\u003cp\u003eTaking statistical and clinical interpretability into consideration, our model eventually identified seven unique trajectory groups with relatively low AIC, BIC, and SABIC values, as well as relatively high log-likelihood ratios (Table 1). The mean posterior probabilities of group membership for the group members were all above 70%, further supporting a great overall fit of the 7-group model (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). The fixed effects of the seven-class longitudinal model are detailed in Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eLatent class growth analysis(LCGA)identified seven distinct dynamic trajectories of MAP within the first 24 hours after ICU admission (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e). Class 1 (n\u0026thinsp;=\u0026thinsp;202, 19.1%) showed moderate MAP values, increasing slightly from approximately 83 mmHg to 86 mmHg. Class 2 (n\u0026thinsp;=\u0026thinsp;112, 10.6%) started with high MAP around 98 mmHg and gradually declined to 88 mmHg. Class 3 (n\u0026thinsp;=\u0026thinsp;59, 5.6%) remained at a consistently elevated level, decreasing modestly from 109 mmHg to 106 mmHg. Class 4 (n\u0026thinsp;=\u0026thinsp;23, 2.2%) exhibited a rising pattern, increasing from 82 mmHg to 108 mmHg. Class 5 (n\u0026thinsp;=\u0026thinsp;174, 16.5%) demonstrated persistently low MAP, remaining stable around 63mmHg. Class 6 (n\u0026thinsp;=\u0026thinsp;56, 5.3%) started from 98 mmHg and showed a sharp decline to approximately 65 mmHg. Class 7 (n\u0026thinsp;=\u0026thinsp;429, 40.7%) represented the largest subgroup, with MAP decreasing slightly from 76 mmHg to 72 mmHg. These heterogeneous trajectories indicate substantial variability in early hemodynamic patterns among patients with sepsis.\u003c/p\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003eComparisons of patient characteristics between trajectory groups\u003c/h2\u003e\n \u003cp\u003eBaseline characteristics differed significantly across the seven MAP trajectory classes (Table 4). Patients in Class 4 were the youngest (median age 58 years), whereas those in Class 5 and Class 6 were the oldest (median age 70.5 and 71.0 years, respectively). Disease severity varied across classes, with the highest APACHE II and SOFA scores observed in Class 5\u0026ndash;7 and the lowest in Class 2\u0026ndash;4. Laboratory tests indicated that patients in Class 5\u0026ndash;7 were more likely to have lower albumin and sodium levels, as well as abnormal coagulation parameters. Inflammatory markers also differed, with Class 7 exhibiting the highest white blood cell count. Arterial blood gas analysis showed variations in pH and lactate levels among groups. Regarding comorbidities, the prevalence of hypertension, acute kidney injury, heart failure, and diabetes differed across trajectory classes. The incidence of delirium at the first assessment also varied significantly: Class 6 had the highest incidence (37.5%), followed by Class 4 (34.8%) and Class 7 (27.5%), whereas Class 5 had the lowest incidence (18.4%), with Class 1 (20.8%) and Class 3 (20.3%) showing relatively lower rates. Patients in Class 5\u0026ndash;7 were more likely to receive vasoactive agents and neuromuscular blockers. No significant differences were observed in gender distribution, pneumonia, myocardial infarction, or ICU 28-day mortality.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003eUnivariate and multivariate analysis\u003c/h2\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eA presents the Kaplan-Meier curves demonstrating the cumulative incidence of delirium among the seven hemodynamic trajectory classes. A significant divergence in delirium risk was observed across the groups (log-rank test, p\u0026thinsp;=\u0026thinsp;0.039). Class 4 and Class 6 exhibited high-risk profiles during the early phase, with a sharp increase in delirium incidence within the first 24 hours after ICU admission, reaching approximately 75% and 60%, respectively\u0026mdash;significantly higher than other classes. In contrast, Class 1 and Class 5 were associated with considerably lower risk, showing gradually rising curves with cumulative incidence rates of 40% and 45% at 120 hours. Classes 2, 3, and 7 demonstrated intermediate risk patterns. Notably, Class 7, despite being the largest subgroup, displayed a persistently increasing risk throughout the mid-to-late observation period. The number-at-risk table confirms sufficient sample sizes during the critical initial phase, supporting the reliability of these findings. These results underscore that early hemodynamic instability, particularly the acute deterioration patterns represented by Class 4 and Class 6, serves as a critical predictor of delirium development in septic patients.\u003c/p\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eB presents the cumulative incidence function (CIF) curves with 7-day mortality as a competing risk. Gray\u0026rsquo;s test indicated statistically significant differences in delirium risk among the seven hemodynamic trajectory classes (p\u0026thinsp;=\u0026thinsp;0.039). Class 4 and Class 6 exhibited high-risk profiles, with a rapid increase in delirium probability during the initial 24 hours. The cumulative incidence of delirium in Class 4 reached approximately 75% by 120 hours, while Class 6 reached approximately 65%. In contrast, Class 1 and Class 5 demonstrated lower and more gradual increases in delirium probability, with 168-hour cumulative incidence rates of approximately 25% and 35%, respectively.\u003c/p\u003e\n \u003cp\u003eCox proportional hazards regression analysis demonstrated significant differences in delirium risk across MAP trajectory classes, as shown in Table \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003e. Notably, Class 4 and Class 6 were consistently associated with markedly increased delirium risk in all models. Using Class 5 as the reference, the hazard ratios for Class 4 ranged from 2.456 to 2.856 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05), and for Class 6 from 2.114 to 2.682 (p\u0026thinsp;\u0026lt;\u0026thinsp;0.01), indicating that patients in these two trajectories were at the highest and most robust risk. In contrast, Classes 2, 3, and 7 showed moderate risk elevation in some models, whereas Class 1 was not significantly associated with delirium.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003eSubgroup analysis\u003c/h2\u003e\n \u003cp\u003eSubgroup analyses were conducted to examine the robustness of the association between MAP trajectory classes and delirium across different clinical characteristics (Table \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003e). No significant interaction effects were observed among subgroups stratified by age, gender, use of sedatives, vasoactive agents, hypertension, or diabetes (all \u003cem\u003eP\u003c/em\u003e for interaction\u0026thinsp;\u0026gt;\u0026thinsp;0.05).\u003c/p\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we analyzed the trajectories of MAP during the first 24 hours after ICU admission in patients with sepsis using data from the MIMIC-IV database. Seven distinct MAP trajectory patterns were identified. The results demonstrated that both the rising (Class 4) and sharply declining (Class 6) MAP trajectories were significantly associated with the occurrence of delirium. This association remained robust after multivariable adjustment and was further confirmed in multiple subgroup analyses. In contrast, patients with persistently low MAP (Class 5) exhibited the lowest incidence of delirium, while those with moderately stable MAP (Class 1) also showed a relatively low risk. The remaining trajectory types were associated with intermediate risk levels. These findings suggest that the development of delirium in sepsis is not solely related to the absolute level of MAP but is also closely linked to the dynamic fluctuations of blood pressure over time.\u003c/p\u003e \u003cp\u003eCerebral autoregulation (CA) is a key mechanism that maintains stable cerebral perfusion, allowing cerebral blood flow to remain relatively constant despite fluctuations in systemic blood pressure. \u003csup\u003e2\u003c/sup\u003eHowever, under critical conditions\u0026mdash;particularly in patients with sepsis, traumatic brain injury, or stroke\u0026mdash;CA function is often impaired, rendering cerebral perfusion pressure passively dependent on mean arterial pressure (MAP). Under such circumstances, large fluctuations in MAP can directly affect cerebral blood flow and oxygen delivery, exacerbating neuronal injury. [13\u0026ndash;16] Previous studies have also shown that increased blood pressure variability (BPV) is a stronger predictor of postoperative delirium than low blood pressure, suggesting that dynamic instability independently contributes to cerebral dysfunction. [17]\u003c/p\u003e \u003cp\u003eExcessive blood pressure fluctuations not only reflect systemic circulatory instability but may also directly damage the vasculature through hemodynamic mechanisms. [15\u0026ndash;18] Studies have shown that prolonged or pronounced blood pressure variability can promote atherosclerosis and increase arterial stiffness, thereby impairing cerebral microvascular perfusion and structural integrity. [18] Chronic high variability in blood pressure may induce cumulative damage throughout the arterial tree down to smaller vessels, further compromising blood flow to arterioles and capillaries, particularly in regions susceptible to ischemic-hypoxic injury, including subcortical white matter and the cerebral cortex. [19] Endothelial cells are highly sensitive to shear stress induced by blood flow; [20] abrupt changes in shear stress can trigger endothelial apoptosis and senescence via PKC ζ, JNK-MAPK, p53, and unfolded protein response pathways. [18,20\u0026ndash;21] In contrast, stable blood flow exerts protective effects through nitric oxide synthase and antioxidant enzyme pathways. [18,22]\u003c/p\u003e \u003cp\u003eBurkhart et al. further highlighted that cerebral perfusion in sepsis-associated encephalopathy (SAE) is regulated by both macrocirculatory and microcirculatory hemodynamics, with blood\u0026ndash;brain barrier (BBB) disruption being a key mechanism. [23] Transcranial Doppler studies by Pfister et al. also confirmed that CA is impaired in SAE patients, rendering cerebral perfusion more susceptible to MAP fluctuations. [24] Collectively, these findings suggest that dynamic blood pressure instability may contribute to cerebral hypoperfusion and delirium through endothelial dysfunction and BBB disruption. In the present study, Class 4 and Class 6 MAP trajectories, representing rapidly rising and sharply declining patterns, likely reflect this impaired cerebral blood flow regulation, which may explain their strong association with increased delirium risk.\u003c/p\u003e \u003cp\u003eThe optimal mean arterial pressure (MAP) target in patients with septic shock remains controversial. Current guidelines recommend maintaining MAP at approximately 65 mmHg; [8] however, the SEPSISPAM trial demonstrated no significant difference in overall mortality between high MAP (80\u0026ndash;85 mmHg) and low MAP (65\u0026ndash;70 mmHg) groups, with the exception that patients with a history of hypertension in the high-target group required less renal replacement therapy. [25] Notably, Deruddre et al. reported that organ perfusion responses vary considerably among individuals once MAP exceeds 65 mmHg. This inter-individual variability is particularly relevant for cerebral perfusion and neurological function. In patients with impaired brain function, a uniform MAP target may result in either excessive or insufficient cerebral perfusion, potentially triggering ischemic or congestive neuronal injury and increasing the risk of delirium. [7] In our study, we observed that patients whose MAP remained stable 63 mmHg exhibited the lowest risk of delirium, while 28-day mortality did not differ significantly compared with other MAP ranges. These findings suggest that fixed MAP targets may overlook individual differences in cerebral autoregulation, thereby amplifying the detrimental effects of blood pressure fluctuations on brain function.\u003c/p\u003e \u003cp\u003eOur findings suggest that the prediction and prevention of delirium should shift from reliance on static blood pressure measurements to dynamic monitoring. Traditional management strategies targeting a single MAP value may overlook blood pressure variability and inter-individual differences in cerebral autoregulation, potentially failing to mitigate the risk of neuronal injury. Blood pressure trajectory analysis provides continuous, dynamic hemodynamic information, facilitating early identification of high-risk patients; among these, Class 4 (rising) and Class 6 (sharply declining) trajectories should be prioritized for close monitoring.\u003c/p\u003e \u003cp\u003eIn hemodynamic management, attention should be given to minimizing blood pressure fluctuations through gradual titration of vasoactive agents, cautious fluid resuscitation, and optimization of sedation strategies, thereby avoiding abrupt changes that may compromise cerebral perfusion. Individualized risk assessment\u0026mdash;taking into account pre-existing hypertension, cerebral autoregulation status, and organ perfusion\u0026mdash;should guide the determination of patient-specific MAP targets.\u003c/p\u003e \u003cp\u003eFuture blood pressure management strategies for septic patients should be more individualized, incorporating CA assessment and near-infrared spectroscopy (NIRS) monitoring to dynamically identify each patient\u0026rsquo;s optimal MAP range. This approach aims to preserve cerebral perfusion while reducing the risk of delirium. Furthermore, integrating blood pressure trajectories with other clinical parameters\u0026mdash;such as oxygen saturation, lactate levels, and neurological monitoring indices\u0026mdash;may facilitate a multidimensional, real-time hemodynamic management strategy, providing more precise neuroprotective interventions for patients with septic shock.\u003c/p\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eStrengths and Limitations\u003c/h2\u003e \u003cp\u003eThe strengths of this study lie in the use of a large-scale cohort of sepsis patients and the application of advanced blood pressure trajectory modeling, which allowed for a comprehensive characterization of early dynamic blood pressure changes. Furthermore, the robustness of the findings was confirmed through multivariable and competing-risk analyses. Nevertheless, several limitations should be acknowledged. First, the retrospective design limits the ability to fully account for potential confounding factors. Second, delirium identification relied on electronic health record documentation, which may be subject to underdiagnosis or misclassification. Third, direct measurements of cerebral perfusion and cerebral autoregulation were not available, and the inferred physiological mechanisms require validation in prospective studies. Additionally, prior studies have suggested that blood pressure variability may be influenced by seasonal and environmental factors, which warrant further investigation in future research.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn conclusion, our study demonstrates that early dynamic MAP trajectories are closely linked to the risk of delirium in patients with sepsis. Specifically, acute hemodynamic deterioration\u0026mdash;manifested as rapidly rising or sharply declining MAP patterns\u0026mdash;can identify individuals at high risk. These findings underscore the clinical importance of continuous, individualized blood pressure monitoring and timely hemodynamic interventions to prevent delirium in critically ill septic patients. Future prospective studies are warranted to validate these observations and to explore targeted strategies for optimizing blood pressure stability in this vulnerable population.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"0\" cellspacing=\"3\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAbbreviation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eFull term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eMAP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eMean arterial pressure\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eICU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eIntensive care unit\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eMIMIC-IV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eMedical Information Mart for Intensive Care IV\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eGBTM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGroup-based trajectory modeling\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eLCGA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eLatent class growth analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eCAM-ICU\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eConfusion Assessment Method for the Intensive Care Unit\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSOFA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSequential Organ Failure Assessment\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eAPACHE II\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAcute Physiology and Chronic Health Evaluation II\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eGCS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGlasgow Coma Scale\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eWBC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eWhite blood cell count\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003ePT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eProthrombin time\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eAPTT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eActivated partial thromboplastin time\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eINR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eInternational normalized ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eAKI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAcute kidney injury\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSBP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSystolic blood pressure\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eDBP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDiastolic blood pressure\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003ePaO₂\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eArterial partial pressure of oxygen\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003ePaCO₂\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eArterial partial pressure of carbon dioxide\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eBPV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eBlood pressure variability\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eCA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCerebral autoregulation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSAE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSepsis-associated encephalopathy\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eBBB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eBlood\u0026ndash;brain barrier\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eNIRS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eNear-infrared spectroscopy\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eKNN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eK-nearest neighbors\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eAIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAkaike information criterion\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eBIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eBayesian information criterion\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSABIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSample-size adjusted Bayesian information criterion\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eCIF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCumulative incidence function\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eHR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eHazard ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eConfidence interval\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eIRB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eInstitutional Review Board\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eSTROBE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eStrengthening the Reporting of Observational Studies in Epidemiology\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study utilized de-identified data from the publicly available MIMIC-IV database. Ethical approval and individual patient consent were not required, as confirmed by the Institutional Review Board of the Beth Israel Deaconess Medical Center. Access to the database was approved after completion of the CITI \u0026ldquo;Data or Specimens Only Research\u0026rdquo; course (Record ID: 68950718).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eClinical trial number\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHuman Ethics and Consent to Participate declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCompeting interest All the authors have disclosed that they do not have any conflicts of interest\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and material\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData supporting the findings of this study are available from the corresponding author upon reasonable request, and all other relevant data are included in the main text.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by the following programs:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003eOutstanding Young Medical Talent Program of the Pudong New Area Health Commission, Shanghai, China (No. PWRq2025-26)\u003c/li\u003e\n \u003cli\u003eDiscipline Leader Training Program of the Pudong New Area Health System, Shanghai (No. PWRd2024-12)\u003c/li\u003e\n \u003cli\u003eThe Project of Key Medical Discipline Group Construction in Shanghai Pudong New Area (No. PWZxq2022-13)\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eThe funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions (CRediT roles)\u003c/strong\u003e\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003eWei Song (H.Z.): Conceptualization; Study design; Data acquisition, analysis, and interpretation; Writing \u0026ndash; original draft; Visualization; Project administration; Final approval; Accountability. \u003cem\u003e(First author)\u003c/em\u003e\u003c/li\u003e\n \u003cli\u003eYijiong Zhang , Jian Wan ,Ye Yuan , Si Peng: Statistical modeling; Figure preparation and editing; Results interpretation; Writing \u0026ndash; review and editing; Funding acquisition. \u003cem\u003e(These authors contributed equally with the first author)\u003c/em\u003e\u003c/li\u003e\n \u003cli\u003eConghui Fan, Song Chen, Qian Zhang, Fei Zhong: Assistance with data analysis; Guidance on manuscript revision.\u003c/li\u003e\n \u003cli\u003eTao Zhang: Software implementation; Data processing; Visualization.\u003c/li\u003e\n \u003cli\u003eXueke Wu: Technical implementation; Data verification.\u003c/li\u003e\n \u003cli\u003eZhen Han: Clinical expertise; Quality assurance.\u003c/li\u003e\n \u003cli\u003eQingzhong Zhao: Manuscript critique; Final approval; Accountability.\u003c/li\u003e\n \u003cli\u003eHaijun Zhu: Critically revised the manuscript, and takes responsibility for the integrity of the work.\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgment\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to express their sincere gratitude to \u003cstrong\u003eWeijian Zhang\u003c/strong\u003e for his valuable software and technical support throughout the course of this study. We are also grateful to \u003cstrong\u003eProf. Jian Wan\u003c/strong\u003e, Director of our department, for his insightful guidance and constructive suggestions on the research design and implementation. In addition, we would like to acknowledge \u003cstrong\u003eProf. Song Chen\u003c/strong\u003e for his generous financial support, which made this work possible.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSong W, et al. Comorbidity of PaO(2)/FiO(2) and nonthyroidal illness syndrome synergistically predicts 28-day mortality in sepsis: A retrospective cohort study from the MIMIC-IV database. Sci Prog. 2025;108:368504251359356.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArmstead WM. Cerebral Blood Flow Autoregulation and Dysautoregulation. Anesthesiol Clin. 2016;34:465\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 to 6 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-infectious-diseases","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"infd","sideBox":"Learn more about [BMC Infectious Diseases](http://bmcinfectdis.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/infd","title":"BMC Infectious Diseases","twitterHandle":"#bmcinfectdis","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"sepsis, delirium, MAP, trajectory analysis, ICU","lastPublishedDoi":"10.21203/rs.3.rs-8759359/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8759359/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eDelirium is a serious complication in septic ICU patients, often related to cerebral hypoperfusion and hemodynamic instability. However, the effect of dynamic blood pressure patterns on delirium risk remains unclear. Therefore, we sought to investigate the association between mean arterial pressure (MAP) trajectories during early ICU stay and the subsequent development of delirium in patients with sepsis.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe retrospectively analyzed 1,055 adults with sepsis from the Medical Information Mart for Intensive Care IV(MIMIC-IV)database. MAP during the first 24 hours after ICU admission was recorded every 2 hours. Group-based trajectory modeling identified distinct MAP patterns. The primary outcome was delirium during ICU stay. Associations between MAP trajectories and delirium risk were evaluated using multivariable Fine\u0026ndash;Gray and Cox models, considering 7-day mortality as a competing event.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eSeven distinct MAP trajectories were identified, showing substantial interindividual variability in early hemodynamic patterns. Classes 4 and 6, characterized by rising and sharply declining MAP, exhibited the highest delirium risk, with cumulative incidences of about 75% and 65% by 120 hours. Using Class 5 (persistently low MAP) as the reference, multivariable analyses showed hazard ratios of 2.456\u0026ndash;2.856 for Class 4 and 2.114\u0026ndash;2.682 for Class 6 (all p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Classes 1 and 5 had the lowest risk, while Classes 2, 3, and 7 showed intermediate risk. Subgroup analyses confirmed consistent associations across demographics and interventions.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eEarly MAP trajectories are strongly associated with delirium risk in sepsis. Acute hemodynamic deterioration, especially rising or sharply declining MAP, identifies high-risk patients.\u003c/p\u003e","manuscriptTitle":"Dynamic Mean Arterial Pressure Trajectories and Risk of Delirium in Patients with Sepsis: A Retrospective Cohort Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-17 14:55:22","doi":"10.21203/rs.3.rs-8759359/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-04-19T06:41:24+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"71806164171312416181814136564048806908","date":"2026-04-09T14:44:47+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-02-12T08:37:54+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-12T08:34:31+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-02-09T07:26:26+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-08T10:01:44+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Infectious Diseases","date":"2026-02-08T09:52:00+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-infectious-diseases","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"infd","sideBox":"Learn more about [BMC Infectious Diseases](http://bmcinfectdis.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/infd","title":"BMC Infectious Diseases","twitterHandle":"#bmcinfectdis","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"2e693beb-d19d-4119-8624-a631a00da440","owner":[],"postedDate":"February 17th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-02-17T14:55:26+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-17 14:55:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8759359","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8759359","identity":"rs-8759359","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}