Identifying High-Risk Patients With Cerebral Venous Thrombosis for Early Intervention Authors: Pichatorn Suppakitjanusant, Myriam Vijil, Chanaka Kahathuduwa, Katie Hodges, Saji Bushnaq, Brittany L. Nordhaus, Mohamad Ezzeldin, Saif Bushnaq Published: 2026 ABSTRACT BACKGROUND:Cerebral venous thrombosis presents with diverse clinical features, making early recognition of patients at risk for deterioration challenging. The cerebral venous thrombosis score was developed by our group and later renamed the CLOTS-EVT score as a mnemonic to facilitate recall and improve ease of clinical use. This 10-point tool, including decreased consciousness (C), low platelets and sodium (L), absence of oral contraceptive use (O), thrombosis burden >3 sinuses (T), seizure (S), and papilledema (E), was developed to predict early deterioration and guide timely venous thrombectomy. We aimed to externally validate this score.METHODS:We retrospectively analyzed adults with cerebral venous thrombosis treated between 2015 and 2017 across 12 centers. Clinical, laboratory, and imaging data were collected, and the CLOTS-EVT score was applied to each patient. Predictive performance was evaluated using sensitivity, specificity, predictive values, and the area under the receiver operating characteristic curve. Correlation with functional outcome at discharge and hospital length of stay was assessed.RESULTS:Eighty patients were analyzed (median age, 38.5 years; 66% female). Headache was the most common symptom; oral contraceptive use was reported in one-third of women. Additional risk factors included pregnancy (7.5%), central nervous system infection (1.25%), malignancy (6.25%), hereditary thrombophilia (3.75%), and smoking (15%). The transverse and sigmoid sinuses were most frequently involved. Anticoagulation was initiated in nearly all patients; 10% underwent endovascular therapy. Overall, 63.8% achieved favorable outcomes, 80% were discharged home, and mortality was 3.8%. In this cohort, 41.3% had decreased consciousness, 47.5% had sodium <139 mEq/L, 38.8% had platelets <225 000/μL, 68.75% were not using oral contraceptives, 25% had thrombosis in >3 sinuses, 20% had seizures, and 22.5% had papilledema. A CLOTS-EVT score ≥4 predicted early deterioration with 82% sensitivity, 61% specificity, and an area under the receiver operating characteristic curve of 0.718. Higher scores correlated with longer hospital stay (r=0.325) and worse functional outcome (r=0.363).CONCLUSIONS:The CLOTS-EVT score demonstrated acceptable performance as a predictor of early deterioration in cerebral venous thrombosis. By combining standard clinical, laboratory, and imaging variables, it provides a practical tool for risk stratification and may inform consideration of venous thrombectomy alongside anticoagulation. CLINICAL PERSPECTIVE What Is New? CLOTS-EVT score is a practical tool to predict early neurological deterioration in cerebral venous thrombosis using routinely available clinical, laboratory, and imaging variables at presentation. CLOTS-EVT score is a practical tool to predict early neurological deterioration in cerebral venous thrombosis using routinely available clinical, laboratory, and imaging variables at presentation. What Are the Clinical Implications? A threshold score of ≥4 effectively identifies high-risk patients who may fail standard anticoagulation, supporting timely escalation of care, including closer monitoring and early neurointerventional evaluation. A threshold score of ≥4 effectively identifies high-risk patients who may fail standard anticoagulation, supporting timely escalation of care, including closer monitoring and early neurointerventional evaluation. Incorporating this score into clinical workflows supports a proactive, tiered management strategy that may optimize timing of intervention and improve outcomes in selected high-risk patients with cerebral venous thrombosis. Incorporating this score into clinical workflows supports a proactive, tiered management strategy that may optimize timing of intervention and improve outcomes in selected high-risk patients with cerebral venous thrombosis. Cerebral venous thrombosis (CVT) is a relatively uncommon cerebrovascular condition characterized by occlusion of cerebral vein or dural venous sinuses. It accounts for 0.5% to 3% of all strokes.1 Its incidence may be as high as 2 per 100 000 adults per year.2 Unlike arterial strokes, the most vulnerable populations include young individuals, particularly women of reproductive age, and patients with a prothrombotic state. Presenting symptoms can range from headaches (90%), seizures (20%–40%), cranial neuropathies, and focal neurological deficits (20%–50%) to encephalopathy and coma (20%).3 Its diverse clinical presentation and the need for confirmation by neuroimaging make it a difficult condition to diagnose, requiring a high level of clinical suspicion. In general, CVT has a favorable outcome, with 80% to 90% of patients achieving functional independence.4 The role of early venous thrombectomy in CVT remains uncertain, and it is not currently considered standard care. Anticoagulation with unfractionated or low-molecular-weight heparin remains the standard initial treatment, even in the presence of intracerebral hemorrhage.5 The objectives of anticoagulation therapy in CVT are to prevent thrombus growth, to facilitate recanalization, and to prevent recurrent venous thromboembolism events. However, 15% of patients with CVT do not respond to medical treatment and have a poor outcome despite early anticoagulation,6 prompting consideration of endovascular therapy (EVT). Observational studies have suggested that intervention within 24 hours may improve outcomes.7 In contrast, the TO-ACT randomized trial (Thrombolysis or Anticoagulation for Cerebral Venous Thrombosis) found no difference in functional independence at 12 months between EVT and standard care.8 However, EVT was associated with fewer postrandomization seizures and a lower frequency of symptomatic intracerebral hemorrhage, suggesting a favorable safety profile. These findings indicate that while EVT may not improve long-term functional outcomes, it could still have a role as rescue therapy in patients who worsen despite anticoagulation. We developed the CVT score9 to identify patients with CVT at high risk of deterioration despite standard therapy, guiding earlier and more aggressive intervention. In this article, we renamed the CVT score to CLOTS-EVT as a mnemonic to facilitate recall and clinical use. Clinical deterioration was defined as an increase of ≥2 National Institutes of Health Stroke Scale score compared with the best neurological status during hospitalization. The 0-10 point score incorporates several clinical and radiographic variables. Decreased level of consciousness (C) contributes 1 point and is defined as a Glasgow Coma Scale score <15, a diagnosis of altered mental status at admission, or documented encephalopathy. Low platelets (platelet count <225 000/μL) and low sodium level (sodium <139 mEq/L; L) each contribute 1 point. The absence of oral contraceptive use (O) contributes 2 points. Thrombosis burden involving >3 sinuses (T) accounts for 1 point, as does the presence of clinical seizure at presentation (S). Papilledema (E) contributes 3 points. These components are used collectively to guide consideration of early venous thrombectomy (Figure 1). It predicted deterioration with 83.2% accuracy, with a sensitivity of 71.4% and a specificity of 76.2%, and scores of ≥5 indicated at least a 50% risk. Overview of the CLOTS-EVT score calculation using clinical, laboratory, and imaging variables. OCP indicates oral contraceptive pill. Methods Objectives The primary goal of this study is to externally validate the CLOTS-EVT score for identifying patients at high risk of deterioration, who may be better candidates for conjunctive venous thrombectomy rather than standard anticoagulation therapy alone. Study Population We conducted a retrospective study of patients treated for CVT between January 2015 and December 2017 across 12 medical centers in Houston, Texas. Institutional review board approval was obtained at each participating site. Inclusion criteria were (1) age ≥18 years and (2) confirmed CVT on neuroimaging with evidence of sinus and cortical vein occlusion. Exclusion criteria were (1) age <18 years and (2) CVT associated with head trauma or skull fractures. Data Collection Baseline demographic data collected at presentation included date of diagnosis, age, sex, and vascular risk factors, including hypertension, diabetes, and prior stroke. Risk factors included pregnancy-related events, central nervous system infection, malignancy, hereditary thrombophilia, and smoking. Clinical variables recorded on admission included headache, altered mental status, motor or sensory deficits, speech disturbance, papilledema, and clinical seizure at onset. Radiological variables included venous infarct and the location of thrombosis. In-hospital data included clinical deterioration, medical versus endovascular treatment, CVT-related mortality, length of hospital stay, functional outcome at discharge using the modified Rankin Scale (mRS) score, and discharge disposition. Long-term management after discharge was also documented, including the type of anticoagulation or antiplatelet therapy. Additional data collected included laboratory results, neuroimaging findings, including computed tomography, magnetic resonance imaging, and digital subtraction angiography, and outpatient follow-up when available. In accordance with the authors’ policies, the data, analytic methods, and study materials used in this study will be available upon reasonable request. Data Analysis With a sample size of 80 patients representing 22 patients who deteriorated during current hospitalization and 58 patients without immediate deterioration, the study was at least 80% powered to detect an area under the receiver operating characteristic curve of at least 0.7 at a significance level of 0.05. Categorical variables were summarized as frequencies and percentages. Continuous variables were expressed as means with SD for parametric distributions or medians with interquartile ranges for nonparametric data. External validation of the CLOTS-EVT score was performed using the pROC and epiR packages in R statistical software (version 4.5.0). The methodology described in our prior publication was used to calculate the CVT total score. Sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristic curve were computed for multiple thresholds of the score to assess discriminatory performance. In addition, linear correlations were explored between the CLOTS-EVT and both the mRS score at discharge and the total length of hospital stay. Missing data were assessed for patterns and mechanisms. Variables with minimal missingness were handled using complete-case analysis. Results A total of 80 patients were analyzed (Table 1). The median age was 38.5 (interquartile range, 28.0–56.3) years, and two-thirds were female, consistent with the known epidemiology of CVT. Headache was the most common presenting symptom, whereas sixth cranial nerve palsy was rare. The mean admission Glasgow Coma Scale score was 14 (SD, 2.36), indicating that most patients presented with preserved consciousness. Regarding the individual components of the CLOTS-EVT score (Table 2), 41.3% of patients presented with decreased consciousness, 47.5% with a sodium level lower than 139 mEq/L, and 38.8% with a platelet count lower than 225 000/μL. Absence of oral contraceptive pill use was observed in 68.75%, while 25% had extensive thrombosis involving >3 sinuses. Seizure activity was documented in 20% and papilledema in 7.5% of patients. Table 1.Baseline Characteristics Baseline Characteristics Table 2.Number and Percentage of Patients Who Present With Each Component of the CLOTS-EVT Score Number and Percentage of Patients Who Present With Each Component of the CLOTS-EVT Score Additional risk factors (Table 1) included pregnancy-related events in 7.5%, central nervous system infection in 1.25%, malignancy in 6.25%, hereditary thrombophilia in 3.75%, and smoking in 15.0%. Thrombus location most frequently involved the transverse and sigmoid sinuses, consistent with typical venous drainage patterns. Anticoagulation was initiated in all patients, while 10% required endovascular intervention, reflecting its role as a rescue strategy for severe cases. Overall (Table 3), 27.5% of patients experienced neurological deterioration during hospitalization; 63.8% achieved favorable outcomes (mRS score 0–2), 80% were discharged home, and mortality occurred in 3.8% of cases. In terms of follow-up, median duration was 90 (interquartile range, 33–174) days, and 34 patients were loss follow-up. Table 3.In-Hospital Outcomes of Patients With Cerebral Venous Thrombosis Stratified by Endovascular Therapy Status In-Hospital Outcomes of Patients With Cerebral Venous Thrombosis Stratified by Endovascular Therapy Status A significant linear correlation was observed between CLOTS-EVT score and mRS score at discharge (r=0.363 [95% CI, 0.155–0.541]; P<0.001). Similarly, a positive linear correlation was noted between CLOTS-EVT score and total length of hospital stay (r=0.325 [95% CI, 0.111–0.511]; P=0.004). Correlation plots did not suggest a significant influence of outliers on these findings. While these are associative rather than causal, they suggest that increasing CLOTS-EVT scores may reflect greater clinical severity. This analysis (Table 4) showed that a CLOTS-EVT score of ≥4 predicted poor outcomes with a sensitivity of 82%, a specificity of 61%, a positive predictive value of 45%, and a negative predictive value of 90%. A threshold of 4 appeared optimal for identifying patients at risk of deterioration, supporting the score’s discriminatory ability, with an area under the receiver operating characteristic curve of 71.8% (Figure 2). Model performance declined when variables were reduced stepwise, further supporting the integrity of the full scoring system. Table 4.Multiple Epidemiological Test Outcomes for Each Threshold of CLOTS-EVT Score in Predicting Early Clinical Deterioration Multiple Epidemiological Test Outcomes for Each Threshold of CLOTS-EVT Score in Predicting Early Clinical Deterioration Receiver operating characteristic curves corresponding to multiple thresholds of the CLOTS-EVT score in predicting early clinical deterioration. Discussion The CLOTS-EVT score9 was developed to identify high-risk patients with CVT at initial presentation who are likely to deteriorate during the acute phase and may require more aggressive care. In this study, we externally validated the score using a data set from a different institution and found that a threshold of ≥4 demonstrated a sensitivity of 82%, a specificity of 61%, and an area under the receiver operating characteristic curve of 71.8%, consistent with our previously proposed cutoff. Importantly, the model is practical for clinical use, as it relies on routinely available symptoms, neurological examination findings, neuroimaging, and laboratory data that can be readily obtained at presentation. The CLOTS-EVT score integrates clinical, laboratory, and radiological variables to predict early deterioration in CVT.9 Clinical predictors include papilledema, seizures, and decreased consciousness. The development and progression of papilledema indicate prolonged intracranial pressure and are associated with a higher risk of permanent visual field deficits and optic atrophy, even with appropriate therapy.10,11 Seizures reflect underlying parenchymal injury, including intracerebral hemorrhage, cerebral edema, infarction, or cortical involvement,12 while decreased consciousness signals extensive brain injury, severe intracranial hypertension, or mass effect, all of which are linked to increased mortality and long-term disability.4 Oral contraceptive pill use identifies a common risk factor in women of reproductive age, whereas CVT in the absence of oral contraceptive pill use may indicate chronic or unprovoked prothrombotic states, such as thrombophilia, malignancy, or autoimmune disease, which are associated with more extensive parenchymal involvement and poorer outcomes.13 Laboratory markers, including hyponatremia and thrombocytopenia, indicate cerebral edema and elevated risk of intracerebral hemorrhage, complicating anticoagulation.4 Radiological predictors, such as clot burden involving ≥3 sinuses and deep venous system thrombosis, correlate with severe intracranial hypertension and rapid neurological decline.14 Overall, the score emphasizes systemic, hematologic, and radiological factors to identify patients at the highest risk of early deterioration and guide timely intervention. When compared with prior interventional trials, clear contrasts can be identified. The TO-ACT trial8 enrolled patients with severe disease, defined by mental status disorder, coma (Glasgow Coma Scale score <9), intracerebral hemorrhage, or deep cerebral venous system thrombosis, thereby capturing individuals already in advanced stages. This may explain the lack of difference in functional independence at 12 months between EVT and medical therapy. However, mounting evidence from observational cohorts7,15 suggests that EVT can be beneficial in carefully selected patients before irreversible parenchymal injury occurs. Early mechanical recanalization may reduce venous congestion, limit infarct expansion, and prevent transtentorial herniation, particularly in those with extensive clot burden or deep venous involvement. The CLOTS-EVT score bridges this gap by identifying patients before the onset of severe neurological deterioration who may still derive benefit from endovascular intervention. By incorporating markers such as seizures, papilledema, and hyponatremia, features reflecting rising intracranial pressure and early parenchymal compromise. The score provides an opportunity to recognize candidates for escalation, while the disease remains in a reversible phase. Integrating the CLOTS-EVT score into clinical workflows could, thus, support a tiered approach to management, where patients at the highest risk (score ≥4) receive early neurointerventional consultation and are prioritized for transfer to centers capable of performing thrombectomy. This proactive strategy may help optimize timing, prevent secondary injury, and improve functional outcomes that were not captured in late-stage intervention trials. Previous scoring systems for CVT have focused primarily on long-term functional outcomes or mortality, in contrast to the present study, which targets early in-hospital deterioration. Prior tools have predicted outcomes such as poor functional status (mRS score, 3–6) or mortality at discharge,16 30 days,17 90 days,18 6 months,19–21 and 1 year.21 A comprehensive meta-analysis of 64 studies over the past 20 years evaluated prognostic factors influencing outcomes in CVT.22 Poor functional outcomes were most strongly associated with ischemic infarction,13,16 impaired consciousness,13,17 cancer, and low absolute lymphocyte count.23 Among imaging findings, intracranial hemorrhage13,15–17 carried the greatest effect size, while ischemic infarction was the most consistent predictor across studies. Inflammatory markers, especially the neutrophil-to-lymphocyte ratio,23 were also correlated with poor outcomes. Several limitations of this study should be considered. First, external validation was conducted with a relatively small cohort, which may limit generalizability. Second, the retrospective design and variability in clinical assessments, laboratory testing, and imaging interpretation may have introduced bias. Third, we employed a complete-only analysis under the assumption that missing data occurred randomly and were not influenced by outcomes. However, the assumption that missingness is completely at random is rarely guaranteed to hold true. Fourth, institutional variation in treatment practices and thresholds for interventions such as mechanical thrombectomy may have influenced outcomes and limited applicability. Prospective studies, ideally randomized controlled trials, are needed to standardize treatment approaches and to further define the clinical utility of the proposed scoring system. Finally, the follow-up period was restricted to short-term outcomes, and the predictive value of the score over longer intervals remains untested. In conclusion, this study provides the first external validation of the CLOTS-EVT score and demonstrates its utility in identifying patients with CVT at high risk of early deterioration. By integrating clinical, laboratory, and radiological factors available at presentation, the score offers a practical tool to guide clinical decision-making regarding which patients should receive early intervention before deterioration occurs. Although the predictive performance was consistent with the original development study, limitations related to sample size and retrospective methodology underscore the need for larger, prospective, multicenter, randomized controlled investigations. Further validation will be essential to establish its role in guiding therapeutic decisions, including the selection and timing of EVT. ARTICLE INFORMATION Sources of Funding None. Disclosures None.