Aneurysmal Subarachnoid Hemorrhage Risk Assessment Model Identifies Patients for Safe Early Discharge at Day 15 – The SAFE-SaHScore

Aneurysmal Subarachnoid Hemorrhage Risk Assessment Model Identifies Patients for Safe Early Discharge at Day 15 – The SAFE-SaHScore | 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 Aneurysmal Subarachnoid Hemorrhage Risk Assessment Model Identifies Patients for Safe Early Discharge at Day 15 – The SAFE-SaHScore Eric E Kennison, Nick M Murray, Dave S Collingridge, Daniel Knox, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5357203/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 28 Mar, 2025 Read the published version in Neurocritical Care → Version 1 posted 5 You are reading this latest preprint version Abstract Background Patients with aneurysmal subarachnoid hemorrhage (aSAH) are often hospitalized for 21 days after aneurysm rupture due to the risk of complications. However, some never experience complications and are unlikely to benefit from a prolonged hospitalization. Objective The aim of this study is to derive a risk assessment model (RAM) using data from the first 14 days of hospitalization to identify low-risk patients for early discharge, at day 15 or after. Methods Patients > 18 years old with an acute aSAH at a Comprehensive Stroke Center from 2017–2024 were included. Baseline demographics, aSAH grading scales, and in-hospital complications requiring intervention were characterized. Complications included: vasospasm, delayed cerebral ischemia (DCI), cerebral salt wasting (CSW), cerebral edema, seizures, arrhythmias, respiratory failure, and hydrocephalus. Binary logistic regression with leave-one-out cross validation (LOOCV) was used to identify an optimal RAM. Results Of 165 patients, the mean Hunt Hess Score (HHS) was 2.5 (SD 1.2), modified Fisher Score (mFS) was 3.1 (SD 1), endovascular therapy was used for aneurysm securement in 73%, and 54.5% experienced complications during days 15–21. In bivariate analyses, days 0–14 variables associated with days 15 + complications were: HHS, mFS, middle cerebral artery (MCA) aneurysm, clinical or radiologic vasospasm, endovascular therapies, intraventricular hemorrhage, hydrocephalus, external ventricular drain (EVD), mechanical ventilation, vasopressors, hypertonic solutions, antiseizure medications, milrinone, and fludrocortisone (all p < 0.05). LOOCV regression for a best fit RAM included 6-variables: S um - of vasopressors, A rtery - MCA aneurysm, F ludrocortisone, E VD, S cale - modified Fisher Score and H unt and Hess Score [ SAFE-SaH ], and had an AUC = 0.90 (0.85–0.95), sensitivity = 0.94, specificity = 0.69, PPV = 79%, and NPV = 91% for predicting complications on day 15+. Conclusions This is the first ever RAM to incorporate clinical data from the first 14 days of hospitalization to identify aSAH patients at low risk for complications after day 14. With 94% sensitivity, the RAM classifies patients who will not have complications and may assist in earlier disposition on day 15 or after. Aneurysmal Subarachnoid Hemorrhage Risk Assessment Model Neurocritical Care Prediction Complication Risk Figures Figure 1 Figure 2 Figure 3 INTRODUCTION Aneurysmal subarachnoid hemorrhage (aSAH) is responsible for 5% of all stroke types and has an incidence of 6.1 per 100,000 person-years. 1 aSAH is a neurological emergency with marked morbidity and mortality that requires close monitoring due to the high rate of complications that follow the initial aneurysm rupture, affecting 8–35% of patients. 2 , 3 These complications include: cerebral vasospasm that can result in secondary stroke and delayed cerebral ischemia (DCI, 5–35% of patients), cerebral salt wasting (CSW, 30% of patients), cerebral edema and hydrocephalus (CE, 5–30% of patients), seizures (8–15% of patients), and cardiac arrhythmias (21% of patients). 1 , 4 – 9 Due to the risk of these complications after aneurysm rupture, patients are generally monitored and treated in the intensive care or acute inpatient setting for 21 days. Complication prevalence is greatest 7–14 days after aneurysm rupture and spontaneously resolves in most patients after 21 days. 1 , 2 Ultimately, some patients never experience complications and are unlikely to benefit from the full 21-day hospitalization. Knowing which patients may and may not require a third week of hospitalization after day 14 is important, as it could prevent unnecessarily long hospital stays and the associated nosocomial and financial burdens. There are no predictive models for delayed complications in patients with an aSAH that incorporate both the admission and the early hospitalization clinical characteristics. Therefore, there is a need for risk assessment models to inform the treatment team if some patients are low risk, and hence safe for discharge after the first 14 days. Existing predictive models for which patients will experience the most severe complication, vasospasm, and the need for prolonged intensive care unit (ICU) monitoring, only consider the initial, admission-based clinical characteristics and head imaging while in the Emergency Department. 10 , 11 There is need to incorporate additional clinical variables from the first part of the patient’s inpatient stay, which may be stronger predictors of development of delayed complications. In this retrospective cohort study, we derived a risk assessment model using common clinical variables from admission as well as day 0–14 of hospitalization to predict the patient-specific risk level for requiring an ICU intervention after day 14. METHODS Study Design This is an institutional review board approved (IRB #1052562) retrospective, single center study at a Comprehensive Stroke Center. Patients were identified from: local REDCap registry, 12 , 13 enterprise data warehouse (EDW), and manual chart review. Clinical, laboratory, medication administration, radiologic, and outcome data of aSAH patients admitted from January 1, 2017, to January 1, 2024, were collected from these databases and verified manually. Patient Selection Patients 18 years or older with a spontaneous SAH were included. Exclusion criteria were: incarcerated status, trauma-related SAH without evidence of aneurysm, perimesencephalic SAH without evidence of aneurysm, the aneurysm was never secured, the patient expired within 14 days of admission, or the data to identify complications could not be obtained from the EMR. Variables Utilized in Risk Assessment Model (RAM), Grouping, and Complications Variables obtained from admission presentation and during the first 14 days of hospitalization which were evaluated for inclusion in the RAM, and included: age, sex, race, medications administered during admission (anti-hypertensives, anti-seizure medications, nimodipine, steroids, vasopressors, inotropes, hyperosmolar therapies), interventions requiring ICU care (external ventricular drain (EVD), mechanical ventilation), grading scale scores (modified Fisher Score (mFS) and Hunt and Hess Score (HHS)), aneurysm location, mechanism in which aneurysm was secured (endovascular versus open neurosurgical), occurrence of clinical or angiographic vasospasm, DCI on CT or MRI, endovascular treatment with verapamil for vasospasm, intraventricular hemorrhage (IVH), hydrocephalus or ventriculoperitoneal (VP) shunt placement, CSF drainage via EVD or lumbar drain (LD), and if a cerebral re-bleed occurred. Full definitions are in Supplementary Table 1. Patients were divided into two groups, based on the presence vs. absence of complications during hospital days 15–21 (detailed in the following paragraph). We then used baseline demographics, hospital admission characteristics, and treatments and interventions within hospital days 0–14 to derive a prediction model to identify patients with vs.without complications in days 15–21. Complications included: EVD placement, mechanical ventilation, endovascular intervention for vasospasm, DCI, CSW, cerebral edema, seizures, and cardiac arrythmias requiring acute treatment. Complete definitions for each variable and complication are in Supplementary Tables 1 and 2. These complications were identified via review of the medications administered during admission and also if documented as such in the daily progress note (e.g., being mechanically ventilated or having an EVD in place). To ensure accuracy of medications administered, all data were validated via manual chart review. Outcomes (dup: abstract ?) The primary outcome was to identify variables on day 0–14 to define patients who had low risk for aSAH related complications on day 15 post bleed and beyond, that is by identifying variables likely to have high sensitivity to detect a complication. This was then used to derive the full risk assessment model (RAM). Statistical Analysis Binary logistic regression with leave-one-out cross validation (LOOCV) was used to identify an optimal model for predicting patients at risk for complications after aSAH. Performance metrics used to identify the best predictive variables included: model variable significance, receiver operating characteristics, area under the curve, accuracy, sensitivity, and specificity. Variables considered in the regression analysis included: sex, age, race, ethnicity, HHS, mFS, vessel of aneurysm location, mechanism of aneurysm securement, transcranial doppler (TCD) velocity elevation, clinical or other angiographic evidence of vasospasm, DCI on CT or MRI, frequency of interventional radiology (IR) interventions for vasospasm treatment, IVH, EVD or LD, VP shunt placement, and re-bleed. Additionally, identified complications that occurred within the first 14 days of admission were also included. Finally, a calibration curve was also created to show predictive ability over the range of probability of complications. The curve shows predicted probability of complication plotted against actual probability of complication. When the apparent line is the same as the 45-degree diagonal line, it indicates predicted and actual probabilities are the same. When the apparent line is above the diagonal line, it indicates the model is underestimating complication. When the apparent line is below the diagonal line, it indicates the model is overestimating complication. RESULTS Patients A total of 295 patients were assessed for eligibility in our risk model derivation and 165 patients were ultimately included in our final analysis after applying exclusion criteria (Figure 1). Mean (SD) age was 56.6 (14.4) years and most patients were white (91.4%) and female (67.9%). The mean HHS and mFS were 2.5 (SD = 1.2) and 3.1 (SD = 1), respectively. Most patients experienced aSAH from the anterior communicating artery (ACoA) (33.9%) or middle cerebral artery (MCA) (18.8%). Mean duration of hospital stay was 20.4 days (SD = 7.1, Table 2). Vasospasm and DCI occurred in 19-63% of patients (depending on clinical versus radiologic classification, or both), hydrocephalus in 50%, CSF diversion was ultimately required in 33.3%, CSW in 36%, and 36% of patients experienced respiratory failure requiring mechanical ventilation. Of the 165 patients, 90 experienced a complication during hospital days 15-21 (Figure 1). Outcomes In bivariate analyses, multiple variables in days 0-14 were associated with complications in days 15-21. These included: HHS (p<0.001), mFS (p<0.001), MCA aneurysm location (p=0.008), elevated TCD velocities (p=0.03), clinical and angiographic evidence of vasospasm (p=0.02 and p<0.001, respectively), IR for vasospasm intervention (p<0.001), IVH (p=0.009), hydrocephalus (p=0.01), placement EVD or LD (p <0.001), mechanical ventilation (p=0.002), use of vasopressors (p<0.001), hypertonic solutions (p<0.001), benzodiazepines or propofol for seizures (p=0.008 and p=0.02), milrinone (p=0.02), verapamil for vasospasm (p<0.001), and fludrocortisone (p<0.001). Complete bivariate analysis can be found in Table 1. We next performed multiple LOOCV regressions for a best fit multivariate model to predict the SAH related complications after day 14 of admission. The final 6 variable model (Model 1) included: number of vasopressors used (OR 5.3, 95% CI 2.0-14.2), MCA aneurysm (OR 7.5, 95% CI 2.0-28.6), fludrocortisone (OR 17.6, 95% CI 5.6-54.9), EVD (OR 5.5, 95% CI 1.8 - 16.9), mFS (OR 1.9, 95% CI 1.1 - 3.4), HHS (OR 1.7, 95% CI 1.1 - 2.7), and had an area under the receiver operator characteristics curve (AUC-ROC) of 0.9 (95% CI, 0.85 – 0.95) (Table 2, Figure 2). The sensitivity and specificity were 0.944 (95% CI, 0.57 – 0.75) and 0.693 (95% CI, 0.78 – 0.90), respectively. The model also generated a positive predictive value (PPV) of 78.7% and a negative predictive value (NPV) of 91.2%. These variables represent the present study name: SAFE-SaH ( S um of vasopressors, A rtery - MCA aneurysm location, F ludrocortisone administration, E VD, S cale - modified Fisher Score and H unt and Hess Score [= SAFE-SaH ]). Clinically, the model can be applied to individual patients to compute their risk for complications. This is done by inputting the 6 aforementioned variables into the derived equation: -5.3662 +1.6621*(sum of vasopressors received, i.e., 1, 2, 3, etc.) + 2.0175*(MCA or not, 1 or 0) + 2.8677*(receipt of fludrocortisone or not, 1 or 0) + 1.6954*(placement of EVD, 1 or 0) + 0.6138*(modified Fisher Score) + 0.5531*(Hunt and Hess Score). The result of this calculation will be in log odds, so this result must then be exponentiated. The output of exponentiating the log odds then can be utilized to calculate the probability of a complication. See the Supplement Spreadsheet File 1, which is a spreadsheet with the formula for easy use of this model application, or alternatively, via this link (https://docs.google.com/spreadsheets/d/17bIqhSNtidZtbmHI2KbwryBi5BFbMdjePBYW_YD2wus/edit?usp=sharing) for the same spreadsheet. Finally, the calibration curve for this best fit model was used to evaluate strength of these predicted probabilities over the spectrum of strength of prediction. The resultant ‘Apparent’ curve is close to the ‘Ideal' curve, though the apparent line is slightly above the Ideal curve when predicted probabilities range from 0 to 0.55, which means the model is slightly underestimating a complication when probabilities are in this range (Figure 3). When the ‘Apparent’ curve is slightly below the diagonal when predicted probabilities are greater than 0.60, the model is slightly overestimating complication when probabilities are in this range. The small area of increase around the predicted probability of 0.50 indicates that the model’s tendency to underestimate complication occurs most often around 0.50. Other multiple variable predictor models using LOOCV had differing, less relevant performance. In sensitivity analyses these models were less suitable for routine clinical use. For example, one of those models, Model 3, was a 7 variable model including: mFS, fludrocortisone, placement of EVD/LD, HHS, use of hypertonic sodium solutions, use of vasopressors, and MCA aneurysm. This model had an AUC-ROC: 0.9 (0.85-0.95), sensitivity=0.93, specificity=0.747, PPV=81.6%, and NPV=90.3%. Another model, Model 2, was a different 6 variable model including the same variables as our presented model (modified Fisher score, Hunt and Hess Score, fludrocortisone, placement of EVD/LD, MCA aneurysm, and use of vasopressors) only differing in the categorization of vasopressor use as a binary option (yes or no) rather than a sum of vasopressor agents used. This model had an AUC-ROC: 0.9 (0.85-0.95), sensitivity=0.92, specificity=0.75, PPV=81.4%, and NPV=88.9%. The AUC-ROC curves and full performance characteristic of these additional models can also be seen in Table 2 and Supplement Figure 1A and 1B. The mortality at hospital discharge of patients who survived day 0-14 was 6.7%. DISCUSSION In this study, we assessed complications of aSAH patients during day 15–21 and then looked back at their clinical parameters day 0–14 to develop a Risk Assessment Model (RAM) to predict risk level of complications at day 15 and beyond. It allows for identification of patients at low risk for complications, which may allow safe, early (after 14 days) discharge. The 6 variable model derived incorporates both the initial emergency department presentation and the early hospitalization characteristics to classify patients’ risk for post aSAH rupture complications. These variables are clinically relevant and easy to identify in routine patient care, and are as follows: S um of vasopressors, A rtery - MCA aneurysm location, F ludrocortisone administration, E VD, S cale - modified Fisher Score and H unt and Hess Score [ SAFE-SaH ]. The resultant RAM, the SAFE-SaH Score, has high reliability and predictive value for identifying risk, which may enable earlier ICU discharge. This is the first study to utilize early hospital characteristics to predict patients at low risk for complications during days 15–21. There are multiple reasons why the selected variables are associated with predictive power for complications from post bleed 15-day and beyond. It is well established that current predictor scales (e.g., mFS and HH) are associated with in-hospital vasospasm and mortality, respectively, hence the radiographic and clinical features on presentation have suitable discrimination power for day 15–21 complication risk. 10 , 14 , 15 Next, the requirement of fludrocortisone and vasopressors align with treating clinically significant cerebral salt wasting and vasospasm, respectively, which adds unique power to our model because these complications are declared only after admission. Finally, placement of CSF diversion devices during the first portion of the patient’s hospitalization demonstrates how interventions for hydrocephalus implicate need for longer hospitalizations given their association with hydrocephalus complications. To our knowledge, there are no prior risk models that use data from the early hospital course to accurately predict patients at low risk for expedited hospital discharge after day 14. Various other models, including active outpatient follow up monitoring, have been explored. One such model, the ‘Fast Track’ identifies aSAH patients who can discharge after hospital day 7 and continue outpatient TCD monitoring if they meet 4 criteria: standard of care aSAH monitoring until day 7, no vasospasm by day 7, no medical comorbidities requiring inpatient care, and availability of caregiver support for 1 week after discharge for required TCD appointments. 16 This model focuses on a single complication (vasospasm) rather than other common complications (e.g., DCI, CSW, seizures, delayed hydrocephalus), which our model incorporates. There are several other risk models that stratify patients with aSAH but these do so only for post-discharge outcomes and complications. These include the: VASOGRADE, SAFIRE, and SAHIT. 15 , 17 , 18 Of note, these models only stratify patients strictly based on their presenting characteristics (World Federation of Neurological Surgeons (WFNS) scores, mFS, aneurysm size, age, etc.), whereas our model incorporates these in addition to the in-hospital characteristics such as medications utilized for early treatment/prevention of complications and the method in which the aneurysm was secured. Our study has several limitations. First, some of the predictors used to derive our model such as scoring scales can be sensitive to user measurement errors. Second, the utilization of medications to define and identify complication treatment in the first 14 hospital days and identify patients with complications days 15–21, may have either underestimated the incidence of the complication, if for instance it was not treated, or over-estimated the incidence if the medication was used for treatment of something other than an acute aSAH complication. However, for the former, a “complication” in a patient not requiring treatment may still mean early discharge is appropriate. This results in reliance on accurate medication charting. These limitations were mitigated by validating via manual chart review to confirm receipt of the medications and the associated indication. Finally, as our RAM was derived in a single center, and both prospective and external validation are needed. Future studies may consider incorporation of additional clinical complication variables and long term tracking of patient outcomes who were discharged early by means of this RAM. CONCLUSION The risk assessment model derived here can predict risk of complications in aSAH patients during day 15–21 post bleed with 94% sensitivity. While further validation is needed, the model may be useful to supplement clinical judgement for timing of discharge in low-risk patients. Declarations Author contributions: EK, NM, DC, DK, and GF developed study conception and design, data acquisition, data analysis, and manuscript writing. DC led data analysis, and also assisted study conception and manuscript writing. EK, NM, and GF assisted with study design and manuscript writing. References Hoh BL, Ko NU, Amin-Hanjani S, et al. 2023 Guideline for the Management of Patients With Aneurysmal Subarachnoid Hemorrhage: A Guideline From the American Heart Association/American Stroke Association. Stroke. 2023;54:e314–70. Treggiari MM, Rabinstein AA, Busl KM, et al. Guidelines for the Neurocritical Care Management of Aneurysmal Subarachnoid Hemorrhage. Neurocrit Care. 2023;39:1–28. Maher M, Schweizer TA, Macdonald RL. Treatment of Spontaneous Subarachnoid Hemorrhage: Guidelines and Gaps. Stroke. 2020;51:1326–32. Osgood ML. Aneurysmal Subarachnoid Hemorrhage: Review of the Pathophysiology and Management Strategies. Curr Neurol Neurosci Rep. 2021;21:50. Suarez JI. Diagnosis and Management of Subarachnoid Hemorrhage. Continuum (Minneap Minn). 2015;21:1263–87. Rowland MJ, Hadjipavlou G, Kelly M, Westbrook J, Pattinson KT. Delayed cerebral ischaemia after subarachnoid haemorrhage: looking beyond vasospasm. Br J Anaesth. 2012;109:315–29. Muehlschlegel S. Subarachnoid Hemorrhage. Continuum (Minneap Minn). 2018;24:1623–57. Ridwan S, Zur B, Kurscheid J, et al. Hyponatremia After Spontaneous Aneurysmal Subarachnoid Hemorrhage-A Prospective Observational Study. World Neurosurg. 2019;129:e538–44. Solenski NJ, Haley EC Jr., Kassell NF, et al. Medical complications of aneurysmal subarachnoid hemorrhage: a report of the multicenter, cooperative aneurysm study. Participants of the Multicenter Cooperative Aneurysm Study. Crit Care Med. 1995;23:1007–17. Claassen J, Bernardini GL, Kreiter K, et al. Effect of cisternal and ventricular blood on risk of delayed cerebral ischemia after subarachnoid hemorrhage: the Fisher scale revisited. Stroke. 2001;32:2012–20. Oliveira Souza NV, Rouanet C, Solla DJF, et al. The Role of VASOGRADE as a Simple Grading Scale to Predict Delayed Cerebral Ischemia and Functional Outcome After Aneurysmal Subarachnoid Hemorrhage. Neurocrit Care. 2023;38:96–104. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inf. 2009;42:377–81. Harris PA, Taylor R, Minor BL, et al. The REDCap consortium: Building an international community of software platform partners. J Biomed Inf. 2019;95:103208. Hunt WE, Hess RM. Surgical risk as related to time of intervention in the repair of intracranial aneurysms. J Neurosurg. 1968;28:14–20. van Donkelaar CE, Bakker NA, Birks J et al. Prediction of Outcome After Aneurysmal Subarachnoid Hemorrhage. Stroke. 2019;50:837 – 44. Collins CI, Hasan TF, Mooney LH, et al. Subarachnoid Hemorrhage Fast Track: A Health Economics and Health Care Redesign Approach for Early Selected Hospital Discharge. Mayo Clin Proc Innov Qual Outcomes. 2020;4:238–48. de Oliveira Manoel AL, Jaja BN, Germans MR, et al. The VASOGRADE: A Simple Grading Scale for Prediction of Delayed Cerebral Ischemia After Subarachnoid Hemorrhage. Stroke. 2015;46:1826–31. Jaja BNR, Saposnik G, Lingsma HF, et al. Development and validation of outcome prediction models for aneurysmal subarachnoid haemorrhage: the SAHIT multinational cohort study. BMJ. 2018;360:j5745. Tables Table 1 and 2 are available in the Supplementary Files section. Supplementary Files Table2.docx SAFESaHScoreCalculationSheet.xlsx SupplementFigure1A.jpg SupplementFigure1B.jpg SupplementaryTable1and2.docx Table1v2.docx Cite Share Download PDF Status: Published Journal Publication published 28 Mar, 2025 Read the published version in Neurocritical Care → Version 1 posted Reviewers agreed at journal 01 Nov, 2024 Reviewers invited by journal 01 Nov, 2024 Editor invited by journal 31 Oct, 2024 Editor assigned by journal 30 Oct, 2024 First submitted to journal 29 Oct, 2024 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-5357203","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":373142535,"identity":"66e68186-b779-4d52-b111-3dfa4ed562a1","order_by":0,"name":"Eric E Kennison","email":"","orcid":"","institution":"Intermountain Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Eric","middleName":"E","lastName":"Kennison","suffix":""},{"id":373142536,"identity":"91770f87-978a-400f-a025-b33f8f854c7e","order_by":1,"name":"Nick M Murray","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1ElEQVRIiWNgGAWjYDACZiCEAsYHQIKHjxQtzAYgLWxE2QMFbBJgkpB6c3bmx0Y3cw7nm7efMav8mmMnw8bA/PDRDTxaLJvZjJNztx22nHMmx+y27LZkoMPYjI1z8GgxOMzDfBioxUCCAahFchszUAsPmzRxWvjfmBVLbqsnTksyWItEjhnjx22HidECdHnutnSglmfF0ozbjvOwMRPyy/nDj6Vzt1kDHZa88ePPbdX2/OzNDx/j04IEOAyYeUA0MyGFCMD+gPEH8apHwSgYBaNgBAEAcBU/M6VCJiwAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0003-3861-0958","institution":"Intermountain Medical Center","correspondingAuthor":true,"prefix":"","firstName":"Nick","middleName":"M","lastName":"Murray","suffix":""},{"id":373142537,"identity":"2d880dad-d967-4fc8-9fa3-98156bded9f7","order_by":2,"name":"Dave S Collingridge","email":"","orcid":"","institution":"Intermountain Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Dave","middleName":"S","lastName":"Collingridge","suffix":""},{"id":373142538,"identity":"c4f35b37-6d96-40a2-bbcd-345287d84ffc","order_by":3,"name":"Daniel Knox","email":"","orcid":"","institution":"Intermountain Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Daniel","middleName":"","lastName":"Knox","suffix":""},{"id":373142539,"identity":"4740c658-6d32-4537-af26-597df3662fea","order_by":4,"name":"Gabriel V Fontaine","email":"","orcid":"","institution":"Intermountain Medical Center","correspondingAuthor":false,"prefix":"","firstName":"Gabriel","middleName":"V","lastName":"Fontaine","suffix":""}],"badges":[],"createdAt":"2024-10-29 23:43:43","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5357203/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5357203/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s12028-025-02236-x","type":"published","date":"2025-03-28T15:57:23+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":70041276,"identity":"f27b79f2-597f-444f-9a1d-71079e50ff1e","added_by":"auto","created_at":"2024-11-27 18:06:01","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":66271,"visible":true,"origin":"","legend":"\u003cp\u003ePatient eligibility and risk model inclusion flow diagram.\u003c/p\u003e","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/267eea686246a7bed7b676c1.jpg"},{"id":70041278,"identity":"bbdb44cd-cca2-4d35-8b79-27c5ddcbef73","added_by":"auto","created_at":"2024-11-27 18:06:01","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":15608,"visible":true,"origin":"","legend":"\u003cp\u003eArea under the receiver operator characteristics curve for the best fit model (#1) for predicting complications day 15 and beyond for aSAH patients.\u003c/p\u003e","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/f831d6966ca56021843326b1.jpg"},{"id":70041772,"identity":"a8d729b9-39b9-474f-9001-01818eff6b15","added_by":"auto","created_at":"2024-11-27 18:14:01","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":41288,"visible":true,"origin":"","legend":"\u003cp\u003eCalibration curve for complications in model 1. This figure indicates predicted probability of complication on the X axis plotted against actual probability of complication on then Y axis. The ‘Apparent’ curve represents the data here, which ideally is as close to the diagonal 45-degree line (the line called ‘Ideal’). The tick marks along the top of the figure show how many patients are in that predicted probability area.\u003c/p\u003e","description":"","filename":"Figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/c73c01e325c6a1828bebfe51.jpg"},{"id":79604962,"identity":"912b69df-a523-4f41-a19a-be08e5ecfb4b","added_by":"auto","created_at":"2025-03-31 16:09:42","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":567181,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/db2f65a6-8627-4355-a8eb-e56268d2c9a4.pdf"},{"id":70041281,"identity":"2c1027bb-8cab-4b8c-8ad7-9fdd66a240f5","added_by":"auto","created_at":"2024-11-27 18:06:01","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":16983,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.docx","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/5598123b689a456a5f6d844e.docx"},{"id":70042535,"identity":"defd5255-ae35-4901-a44a-a8a71c8c15af","added_by":"auto","created_at":"2024-11-27 18:30:01","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":10783,"visible":true,"origin":"","legend":"","description":"","filename":"SAFESaHScoreCalculationSheet.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/fcf49cdb0134b70021797e17.xlsx"},{"id":70041284,"identity":"11a28117-fbbf-4451-b45d-29ef8584a42b","added_by":"auto","created_at":"2024-11-27 18:06:01","extension":"jpg","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":12572,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementFigure1A.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/f16a690780d051620dd9a8ad.jpg"},{"id":70042095,"identity":"cc76b459-e319-41e6-87bb-6667f10e56ad","added_by":"auto","created_at":"2024-11-27 18:22:01","extension":"jpg","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":12591,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementFigure1B.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/8fe4f9bdedbe685aa79f9541.jpg"},{"id":70041776,"identity":"63ec150c-3c15-459b-b4b6-505331db4f58","added_by":"auto","created_at":"2024-11-27 18:14:01","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":18782,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable1and2.docx","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/d2f247196c2ee46d17951ad9.docx"},{"id":70041775,"identity":"49683fa0-9c7a-4e1c-b811-b267e5a690e7","added_by":"auto","created_at":"2024-11-27 18:14:01","extension":"docx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":37565,"visible":true,"origin":"","legend":"","description":"","filename":"Table1v2.docx","url":"https://assets-eu.researchsquare.com/files/rs-5357203/v1/cff69248991e0bda352cf3ab.docx"}],"financialInterests":"","formattedTitle":"Aneurysmal Subarachnoid Hemorrhage Risk Assessment Model Identifies Patients for Safe Early Discharge at Day 15 – The SAFE-SaHScore","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eAneurysmal subarachnoid hemorrhage (aSAH) is responsible for 5% of all stroke types and has an incidence of 6.1 per 100,000 person-years.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e aSAH is a neurological emergency with marked morbidity and mortality that requires close monitoring due to the high rate of complications that follow the initial aneurysm rupture, affecting 8\u0026ndash;35% of patients.\u003csup\u003e\u003cem\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/em\u003e\u003c/sup\u003e These complications include: cerebral vasospasm that can result in secondary stroke and delayed cerebral ischemia (DCI, 5\u0026ndash;35% of patients), cerebral salt wasting (CSW, 30% of patients), cerebral edema and hydrocephalus (CE, 5\u0026ndash;30% of patients), seizures (8\u0026ndash;15% of patients), and cardiac arrhythmias (21% of patients).\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan additionalcitationids=\"CR5 CR6 CR7 CR8\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eDue to the risk of these complications after aneurysm rupture, patients are generally monitored and treated in the intensive care or acute inpatient setting for 21 days. Complication prevalence is greatest 7\u0026ndash;14 days after aneurysm rupture and spontaneously resolves in most patients after 21 days.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e Ultimately, some patients never experience complications and are unlikely to benefit from the full 21-day hospitalization. Knowing which patients may and may not require a third week of hospitalization after day 14 is important, as it could prevent unnecessarily long hospital stays and the associated nosocomial and financial burdens.\u003c/p\u003e \u003cp\u003eThere are no predictive models for delayed complications in patients with an aSAH that incorporate both the admission and the early hospitalization clinical characteristics. Therefore, there is a need for risk assessment models to inform the treatment team if some patients are low risk, and hence safe for discharge after the first 14 days. Existing predictive models for which patients will experience the most severe complication, vasospasm, and the need for prolonged intensive care unit (ICU) monitoring, only consider the initial, admission-based clinical characteristics and head imaging while in the Emergency Department.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e There is need to incorporate additional clinical variables from the first part of the patient\u0026rsquo;s inpatient stay, which may be stronger predictors of development of delayed complications.\u003c/p\u003e \u003cp\u003eIn this retrospective cohort study, we derived a risk assessment model using common clinical variables from admission as well as day 0\u0026ndash;14 of hospitalization to predict the patient-specific risk level for requiring an ICU intervention after day 14.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Design\u003c/h2\u003e \u003cp\u003e This is an institutional review board approved (IRB #1052562) retrospective, single center study at a Comprehensive Stroke Center. Patients were identified from: local REDCap registry,\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e enterprise data warehouse (EDW), and manual chart review. Clinical, laboratory, medication administration, radiologic, and outcome data of aSAH patients admitted from January 1, 2017, to January 1, 2024, were collected from these databases and verified manually.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003ePatient Selection\u003c/h3\u003e\n\u003cp\u003ePatients 18 years or older with a spontaneous SAH were included. Exclusion criteria were: incarcerated status, trauma-related SAH without evidence of aneurysm, perimesencephalic SAH without evidence of aneurysm, the aneurysm was never secured, the patient expired within 14 days of admission, or the data to identify complications could not be obtained from the EMR.\u003c/p\u003e\n\u003ch3\u003eVariables Utilized in Risk Assessment Model (RAM), Grouping, and Complications\u003c/h3\u003e\n\u003cp\u003eVariables obtained from admission presentation and during the first 14 days of hospitalization which were evaluated for inclusion in the RAM, and included: age, sex, race, medications administered during admission (anti-hypertensives, anti-seizure medications, nimodipine, steroids, vasopressors, inotropes, hyperosmolar therapies), interventions requiring ICU care (external ventricular drain (EVD), mechanical ventilation), grading scale scores (modified Fisher Score (mFS) and Hunt and Hess Score (HHS)), aneurysm location, mechanism in which aneurysm was secured (endovascular versus open neurosurgical), occurrence of clinical or angiographic vasospasm, DCI on CT or MRI, endovascular treatment with verapamil for vasospasm, intraventricular hemorrhage (IVH), hydrocephalus or ventriculoperitoneal (VP) shunt placement, CSF drainage via EVD or lumbar drain (LD), and if a cerebral re-bleed occurred. Full definitions are in Supplementary Table\u0026nbsp;1.\u003c/p\u003e \u003cp\u003ePatients were divided into two groups, based on the presence vs. absence of complications during hospital days 15\u0026ndash;21 (detailed in the following paragraph). We then used baseline demographics, hospital admission characteristics, and treatments and interventions within hospital days 0\u0026ndash;14 to derive a prediction model to identify patients with vs.without complications in days 15\u0026ndash;21.\u003c/p\u003e \u003cp\u003eComplications included: EVD placement, mechanical ventilation, endovascular intervention for vasospasm, DCI, CSW, cerebral edema, seizures, and cardiac arrythmias requiring acute treatment. Complete definitions for each variable and complication are in Supplementary Tables\u0026nbsp;1 and 2. These complications were identified via review of the medications administered during admission and also if documented as such in the daily progress note (e.g., being mechanically ventilated or having an EVD in place). To ensure accuracy of medications administered, all data were validated via manual chart review.\u003c/p\u003e\n\u003ch3\u003eOutcomes (dup: abstract ?)\u003c/h3\u003e\n\u003cp\u003eThe primary outcome was to identify variables on day 0\u0026ndash;14 to define patients who had low risk for aSAH related complications on day 15 post bleed and beyond, that is by identifying variables likely to have high sensitivity to detect a complication. This was then used to derive the full risk assessment model (RAM).\u003c/p\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eStatistical Analysis\u003c/h2\u003e \u003cp\u003eBinary logistic regression with leave-one-out cross validation (LOOCV) was used to identify an optimal model for predicting patients at risk for complications after aSAH. Performance metrics used to identify the best predictive variables included: model variable significance, receiver operating characteristics, area under the curve, accuracy, sensitivity, and specificity.\u003c/p\u003e \u003cp\u003eVariables considered in the regression analysis included: sex, age, race, ethnicity, HHS, mFS, vessel of aneurysm location, mechanism of aneurysm securement, transcranial doppler (TCD) velocity elevation, clinical or other angiographic evidence of vasospasm, DCI on CT or MRI, frequency of interventional radiology (IR) interventions for vasospasm treatment, IVH, EVD or LD, VP shunt placement, and re-bleed. Additionally, identified complications that occurred within the first 14 days of admission were also included.\u003c/p\u003e \u003cp\u003eFinally, a calibration curve was also created to show predictive ability over the range of probability of complications. The curve shows predicted probability of complication plotted against actual probability of complication. When the apparent line is the same as the 45-degree diagonal line, it indicates predicted and actual probabilities are the same. When the apparent line is above the diagonal line, it indicates the model is underestimating complication. When the apparent line is below the diagonal line, it indicates the model is overestimating complication.\u003c/p\u003e \u003c/div\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cu\u003ePatients\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eA total of 295 patients were assessed for eligibility in our risk model derivation and 165 patients were ultimately included in our final analysis after applying exclusion criteria (Figure 1). Mean (SD) age was 56.6 (14.4) years and most patients were white (91.4%) and female (67.9%). The mean HHS and mFS were 2.5 (SD = 1.2) and 3.1 (SD = 1), respectively. Most patients experienced aSAH from the anterior communicating artery (ACoA) (33.9%) or middle cerebral artery (MCA) (18.8%). Mean duration of hospital stay was 20.4 days (SD = 7.1, Table 2). Vasospasm and DCI occurred in 19-63% of patients (depending on clinical versus radiologic classification, or both), hydrocephalus in 50%, CSF diversion was ultimately required in 33.3%, CSW in 36%, and 36% of patients experienced respiratory failure requiring mechanical ventilation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOf the 165 patients, 90 experienced a complication during hospital days 15-21 (Figure 1).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cu\u003eOutcomes\u003c/u\u003e\u003c/p\u003e\n\u003cp\u003eIn bivariate analyses, multiple variables in days 0-14 were associated with complications in days 15-21. These included: HHS (p\u0026lt;0.001), mFS (p\u0026lt;0.001), MCA aneurysm location (p=0.008), elevated TCD velocities (p=0.03), clinical and angiographic evidence of vasospasm (p=0.02 and p\u0026lt;0.001, respectively), IR for vasospasm intervention (p\u0026lt;0.001), IVH (p=0.009), hydrocephalus (p=0.01), placement EVD or LD (p \u0026lt;0.001), mechanical ventilation (p=0.002), use of vasopressors (p\u0026lt;0.001), hypertonic solutions (p\u0026lt;0.001), benzodiazepines or propofol for seizures (p=0.008 and p=0.02), milrinone (p=0.02), verapamil for vasospasm (p\u0026lt;0.001), and fludrocortisone (p\u0026lt;0.001). Complete bivariate analysis can be found in Table 1.\u003c/p\u003e\n\u003cp\u003eWe next performed multiple LOOCV regressions for a best fit multivariate model to predict the SAH related complications after day 14 of admission. The final 6 variable model (Model 1) included: number of vasopressors used (OR 5.3, 95% CI 2.0-14.2), MCA aneurysm (OR 7.5, 95% CI 2.0-28.6), fludrocortisone (OR 17.6, 95% CI 5.6-54.9), EVD (OR 5.5, 95% CI 1.8 - 16.9), mFS (OR 1.9, 95% CI 1.1 - 3.4), HHS (OR 1.7, 95% CI 1.1 - 2.7), and had an area under the receiver operator characteristics curve (AUC-ROC) of 0.9 (95% CI, 0.85 \u0026ndash; 0.95) (Table 2, Figure 2). The sensitivity and specificity were 0.944 (95% CI, 0.57 \u0026ndash; 0.75) and 0.693 (95% CI, 0.78 \u0026ndash; 0.90), respectively. The model also generated a positive predictive value (PPV) of 78.7% and a negative predictive value (NPV) of 91.2%. These variables represent the present study name: \u003cstrong\u003eSAFE-SaH\u003c/strong\u003e (\u003cstrong\u003eS\u003c/strong\u003eum of vasopressors, \u003cstrong\u003eA\u003c/strong\u003ertery - MCA aneurysm location, \u003cstrong\u003eF\u003c/strong\u003eludrocortisone administration, \u003cstrong\u003eE\u003c/strong\u003eVD, \u003cstrong\u003eS\u003c/strong\u003ecale - modified Fisher Score and \u003cstrong\u003eH\u003c/strong\u003eunt and Hess Score [= \u003cstrong\u003eSAFE-SaH\u003c/strong\u003e]).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; Clinically, the model can be applied to individual patients to compute their risk for complications. This is done by inputting the 6 aforementioned variables into the derived equation: -5.3662 +1.6621*(sum of vasopressors received, i.e., 1, 2, 3, etc.) + 2.0175*(MCA or not, 1 or 0) \u0026nbsp;+ 2.8677*(receipt of fludrocortisone or not, 1 or 0) + 1.6954*(placement of EVD, 1 or 0) + 0.6138*(modified Fisher Score) + 0.5531*(Hunt and Hess Score). The result of this calculation will be in log odds, so this result must then be exponentiated. The output of exponentiating the log odds then can be utilized to calculate the probability of a complication. See the Supplement Spreadsheet File 1, which is a spreadsheet with the formula for easy use of this model application, or alternatively, via this link (https://docs.google.com/spreadsheets/d/17bIqhSNtidZtbmHI2KbwryBi5BFbMdjePBYW_YD2wus/edit?usp=sharing) for the same spreadsheet.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;Finally, the calibration curve for this best fit model was used to evaluate strength of these predicted probabilities over the spectrum of strength of prediction. The resultant\u0026nbsp;\u0026lsquo;Apparent\u0026rsquo; curve is close to the \u0026lsquo;Ideal\u0026apos; curve, though the apparent line is slightly above the Ideal curve when predicted probabilities range from 0 to 0.55, which means the model is slightly underestimating a complication when probabilities are in this range (Figure 3). When the \u0026lsquo;Apparent\u0026rsquo; curve is slightly below the diagonal when predicted probabilities are greater than 0.60, the model is slightly overestimating complication when probabilities are in this range. The small area of increase around the predicted probability of 0.50 indicates that the model\u0026rsquo;s tendency to underestimate complication occurs most often around 0.50.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eOther multiple variable predictor models using LOOCV had differing, less relevant performance. In sensitivity analyses these models were less suitable for routine clinical use. For example, one of those models, Model 3, was a 7 variable model including: mFS, fludrocortisone, placement of EVD/LD, HHS, use of hypertonic sodium solutions, use of vasopressors, and MCA aneurysm. This model had an AUC-ROC: 0.9 (0.85-0.95), sensitivity=0.93, specificity=0.747, PPV=81.6%, and NPV=90.3%. Another model, Model 2, was a different 6 variable model including the same variables as our presented model (modified Fisher score, Hunt and Hess Score, fludrocortisone, placement of EVD/LD, MCA aneurysm, and use of vasopressors) only differing in the categorization of vasopressor use as a binary option (yes or no) rather than a sum of vasopressor agents used. This model had an AUC-ROC: 0.9 (0.85-0.95), sensitivity=0.92, specificity=0.75, PPV=81.4%, and NPV=88.9%. The AUC-ROC curves and full performance characteristic of these additional models can also be seen in Table 2 and Supplement Figure 1A and 1B.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe mortality at hospital discharge of patients who survived day 0-14 was 6.7%.\u0026nbsp;\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn this study, we assessed complications of aSAH patients during day 15\u0026ndash;21 and then looked back at their clinical parameters day 0\u0026ndash;14 to develop a Risk Assessment Model (RAM) to predict risk level of complications at day 15 and beyond. It allows for identification of patients at low risk for complications, which may allow safe, early (after 14 days) discharge.\u003c/p\u003e \u003cp\u003eThe 6 variable model derived incorporates both the initial emergency department presentation and the early hospitalization characteristics to classify patients\u0026rsquo; risk for post aSAH rupture complications. These variables are clinically relevant and easy to identify in routine patient care, and are as follows: \u003cb\u003eS\u003c/b\u003eum of vasopressors, \u003cb\u003eA\u003c/b\u003ertery - MCA aneurysm location, \u003cb\u003eF\u003c/b\u003eludrocortisone administration, \u003cb\u003eE\u003c/b\u003eVD, \u003cb\u003eS\u003c/b\u003ecale - modified Fisher Score and \u003cb\u003eH\u003c/b\u003eunt and Hess Score [\u003cb\u003eSAFE-SaH\u003c/b\u003e]. The resultant RAM, the SAFE-SaH Score, has high reliability and predictive value for identifying risk, which may enable earlier ICU discharge. This is the first study to utilize early hospital characteristics to predict patients at low risk for complications during days 15\u0026ndash;21.\u003c/p\u003e \u003cp\u003eThere are multiple reasons why the selected variables are associated with predictive power for complications from post bleed 15-day and beyond. It is well established that current predictor scales (e.g., mFS and HH) are associated with in-hospital vasospasm and mortality, respectively, hence the radiographic and clinical features on presentation have suitable discrimination power for day 15\u0026ndash;21 complication risk.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e Next, the requirement of fludrocortisone and vasopressors align with treating clinically significant cerebral salt wasting and vasospasm, respectively, which adds unique power to our model because these complications are declared only after admission. Finally, placement of CSF diversion devices during the first portion of the patient\u0026rsquo;s hospitalization demonstrates how interventions for hydrocephalus implicate need for longer hospitalizations given their association with hydrocephalus complications.\u003c/p\u003e \u003cp\u003eTo our knowledge, there are no prior risk models that use data from the early hospital course to accurately predict patients at low risk for expedited hospital discharge after day 14. Various other models, including active outpatient follow up monitoring, have been explored. One such model, the \u0026lsquo;Fast Track\u0026rsquo; identifies aSAH patients who can discharge after hospital day 7 and continue outpatient TCD monitoring if they meet 4 criteria: standard of care aSAH monitoring until day 7, no vasospasm by day 7, no medical comorbidities requiring inpatient care, and availability of caregiver support for 1 week after discharge for required TCD appointments.\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e This model focuses on a single complication (vasospasm) rather than other common complications (e.g., DCI, CSW, seizures, delayed hydrocephalus), which our model incorporates.\u003c/p\u003e \u003cp\u003eThere are several other risk models that stratify patients with aSAH but these do so only for post-discharge outcomes and complications. These include the: VASOGRADE, SAFIRE, and SAHIT.\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e Of note, these models only stratify patients strictly based on their presenting characteristics (World Federation of Neurological Surgeons (WFNS) scores, mFS, aneurysm size, age, etc.), whereas our model incorporates these in addition to the in-hospital characteristics such as medications utilized for early treatment/prevention of complications and the method in which the aneurysm was secured.\u003c/p\u003e \u003cp\u003eOur study has several limitations. First, some of the predictors used to derive our model such as scoring scales can be sensitive to user measurement errors. Second, the utilization of medications to define and identify complication treatment in the first 14 hospital days and identify patients with complications days 15\u0026ndash;21, may have either underestimated the incidence of the complication, if for instance it was not treated, or over-estimated the incidence if the medication was used for treatment of something other than an acute aSAH complication. However, for the former, a \u0026ldquo;complication\u0026rdquo; in a patient not requiring treatment may still mean early discharge is appropriate. This results in reliance on accurate medication charting. These limitations were mitigated by validating via manual chart review to confirm receipt of the medications and the associated indication.\u003c/p\u003e \u003cp\u003eFinally, as our RAM was derived in a single center, and both prospective and external validation are needed. Future studies may consider incorporation of additional clinical complication variables and long term tracking of patient outcomes who were discharged early by means of this RAM.\u003c/p\u003e"},{"header":"CONCLUSION","content":"\u003cp\u003eThe risk assessment model derived here can predict risk of complications in aSAH patients during day 15\u0026ndash;21 post bleed with 94% sensitivity. While further validation is needed, the model may be useful to supplement clinical judgement for timing of discharge in low-risk patients.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor contributions:\u003c/h2\u003e \u003cp\u003eEK, NM, DC, DK, and GF developed study conception and design, data acquisition, data analysis, and manuscript writing. DC led data analysis, and also assisted study conception and manuscript writing. EK, NM, and GF assisted with study design and manuscript writing.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eHoh BL, Ko NU, Amin-Hanjani S, et al. 2023 Guideline for the Management of Patients With Aneurysmal Subarachnoid Hemorrhage: A Guideline From the American Heart Association/American Stroke Association. Stroke. 2023;54:e314\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTreggiari MM, Rabinstein AA, Busl KM, et al. Guidelines for the Neurocritical Care Management of Aneurysmal Subarachnoid Hemorrhage. Neurocrit Care. 2023;39:1\u0026ndash;28.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaher M, Schweizer TA, Macdonald RL. Treatment of Spontaneous Subarachnoid Hemorrhage: Guidelines and Gaps. Stroke. 2020;51:1326\u0026ndash;32.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOsgood ML. Aneurysmal Subarachnoid Hemorrhage: Review of the Pathophysiology and Management Strategies. Curr Neurol Neurosci Rep. 2021;21:50.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuarez JI. Diagnosis and Management of Subarachnoid Hemorrhage. Continuum (Minneap Minn). 2015;21:1263\u0026ndash;87.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRowland MJ, Hadjipavlou G, Kelly M, Westbrook J, Pattinson KT. Delayed cerebral ischaemia after subarachnoid haemorrhage: looking beyond vasospasm. Br J Anaesth. 2012;109:315\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMuehlschlegel S. Subarachnoid Hemorrhage. Continuum (Minneap Minn). 2018;24:1623\u0026ndash;57.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRidwan S, Zur B, Kurscheid J, et al. Hyponatremia After Spontaneous Aneurysmal Subarachnoid Hemorrhage-A Prospective Observational Study. World Neurosurg. 2019;129:e538\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSolenski NJ, Haley EC Jr., Kassell NF, et al. Medical complications of aneurysmal subarachnoid hemorrhage: a report of the multicenter, cooperative aneurysm study. Participants of the Multicenter Cooperative Aneurysm Study. Crit Care Med. 1995;23:1007\u0026ndash;17.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClaassen J, Bernardini GL, Kreiter K, et al. Effect of cisternal and ventricular blood on risk of delayed cerebral ischemia after subarachnoid hemorrhage: the Fisher scale revisited. Stroke. 2001;32:2012\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOliveira Souza NV, Rouanet C, Solla DJF, et al. The Role of VASOGRADE as a Simple Grading Scale to Predict Delayed Cerebral Ischemia and Functional Outcome After Aneurysmal Subarachnoid Hemorrhage. Neurocrit Care. 2023;38:96\u0026ndash;104.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inf. 2009;42:377\u0026ndash;81.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarris PA, Taylor R, Minor BL, et al. The REDCap consortium: Building an international community of software platform partners. J Biomed Inf. 2019;95:103208.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHunt WE, Hess RM. Surgical risk as related to time of intervention in the repair of intracranial aneurysms. J Neurosurg. 1968;28:14\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Donkelaar CE, Bakker NA, Birks J et al. Prediction of Outcome After Aneurysmal Subarachnoid Hemorrhage. Stroke. 2019;50:837\u0026thinsp;\u0026ndash;\u0026thinsp;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollins CI, Hasan TF, Mooney LH, et al. Subarachnoid Hemorrhage Fast Track: A Health Economics and Health Care Redesign Approach for Early Selected Hospital Discharge. Mayo Clin Proc Innov Qual Outcomes. 2020;4:238\u0026ndash;48.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Oliveira Manoel AL, Jaja BN, Germans MR, et al. The VASOGRADE: A Simple Grading Scale for Prediction of Delayed Cerebral Ischemia After Subarachnoid Hemorrhage. Stroke. 2015;46:1826\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJaja BNR, Saposnik G, Lingsma HF, et al. Development and validation of outcome prediction models for aneurysmal subarachnoid haemorrhage: the SAHIT multinational cohort study. BMJ. 2018;360:j5745.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 and 2 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":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"neurocritical-care","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"neca","sideBox":"Learn more about [Neurocritical Care](http://link.springer.com/journal/12028)","snPcode":"12028","submissionUrl":"https://www.editorialmanager.com/neca/default2.aspx","title":"Neurocritical Care","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Aneurysmal Subarachnoid Hemorrhage, Risk Assessment Model, Neurocritical Care, Prediction, Complication Risk ","lastPublishedDoi":"10.21203/rs.3.rs-5357203/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5357203/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e \u003cp\u003ePatients with aneurysmal subarachnoid hemorrhage (aSAH) are often hospitalized for 21 days after aneurysm rupture due to the risk of complications. However, some never experience complications and are unlikely to benefit from a prolonged hospitalization.\u003c/p\u003e\u003cp\u003e\u003cb\u003eObjective\u003c/b\u003e\u003c/p\u003e \u003cp\u003eThe aim of this study is to derive a risk assessment model (RAM) using data from the first 14 days of hospitalization to identify low-risk patients for early discharge, at day 15 or after.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e \u003cp\u003ePatients\u0026thinsp;\u0026gt;\u0026thinsp;18 years old with an acute aSAH at a Comprehensive Stroke Center from 2017\u0026ndash;2024 were included. Baseline demographics, aSAH grading scales, and in-hospital complications requiring intervention were characterized. Complications included: vasospasm, delayed cerebral ischemia (DCI), cerebral salt wasting (CSW), cerebral edema, seizures, arrhythmias, respiratory failure, and hydrocephalus. Binary logistic regression with leave-one-out cross validation (LOOCV) was used to identify an optimal RAM.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e \u003cp\u003eOf 165 patients, the mean Hunt Hess Score (HHS) was 2.5 (SD 1.2), modified Fisher Score (mFS) was 3.1 (SD 1), endovascular therapy was used for aneurysm securement in 73%, and 54.5% experienced complications during days 15\u0026ndash;21. In bivariate analyses, days 0\u0026ndash;14 variables associated with days 15\u0026thinsp;+\u0026thinsp;complications were: HHS, mFS, middle cerebral artery (MCA) aneurysm, clinical or radiologic vasospasm, endovascular therapies, intraventricular hemorrhage, hydrocephalus, external ventricular drain (EVD), mechanical ventilation, vasopressors, hypertonic solutions, antiseizure medications, milrinone, and fludrocortisone (all p\u0026thinsp;\u0026lt;\u0026thinsp;0.05). LOOCV regression for a best fit RAM included 6-variables: \u003cb\u003eS\u003c/b\u003eum - of vasopressors, \u003cb\u003eA\u003c/b\u003ertery - MCA aneurysm, \u003cb\u003eF\u003c/b\u003eludrocortisone, \u003cb\u003eE\u003c/b\u003eVD, \u003cb\u003eS\u003c/b\u003ecale - modified Fisher Score and \u003cb\u003eH\u003c/b\u003eunt and Hess Score [\u003cb\u003eSAFE-SaH\u003c/b\u003e], and had an AUC\u0026thinsp;=\u0026thinsp;0.90 (0.85\u0026ndash;0.95), sensitivity\u0026thinsp;=\u0026thinsp;0.94, specificity\u0026thinsp;=\u0026thinsp;0.69, PPV\u0026thinsp;=\u0026thinsp;79%, and NPV\u0026thinsp;=\u0026thinsp;91% for predicting complications on day 15+.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions\u003c/b\u003e\u003c/p\u003e \u003cp\u003eThis is the first ever RAM to incorporate clinical data from the first 14 days of hospitalization to identify aSAH patients at low risk for complications after day 14. With 94% sensitivity, the RAM classifies patients who will not have complications and may assist in earlier disposition on day 15 or after.\u003c/p\u003e","manuscriptTitle":"Aneurysmal Subarachnoid Hemorrhage Risk Assessment Model Identifies Patients for Safe Early Discharge at Day 15 – The SAFE-SaHScore","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-11-27 18:05:56","doi":"10.21203/rs.3.rs-5357203/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2024-11-01T22:26:42+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-11-01T22:20:46+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"Neurocritical Care","date":"2024-10-31T22:23:33+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-10-30T18:44:50+00:00","index":"","fulltext":""},{"type":"submitted","content":"Neurocritical Care","date":"2024-10-29T19:42:46+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"neurocritical-care","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"neca","sideBox":"Learn more about [Neurocritical Care](http://link.springer.com/journal/12028)","snPcode":"12028","submissionUrl":"https://www.editorialmanager.com/neca/default2.aspx","title":"Neurocritical Care","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"f49c1e8c-e94d-4ca9-a5f6-c395c5f60af3","owner":[],"postedDate":"November 27th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-03-31T16:04:54+00:00","versionOfRecord":{"articleIdentity":"rs-5357203","link":"https://doi.org/10.1007/s12028-025-02236-x","journal":{"identity":"neurocritical-care","isVorOnly":false,"title":"Neurocritical Care"},"publishedOn":"2025-03-28 15:57:23","publishedOnDateReadable":"March 28th, 2025"},"versionCreatedAt":"2024-11-27 18:05:56","video":"","vorDoi":"10.1007/s12028-025-02236-x","vorDoiUrl":"https://doi.org/10.1007/s12028-025-02236-x","workflowStages":[]},"version":"v1","identity":"rs-5357203","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5357203","identity":"rs-5357203","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}