Introduction

Mechanical prosthetic valve thrombosis remains one of the most serious and life‑threatening complications following valve replacement surgery, despite substantial advances in prosthetic design and anticoagulation management. The condition is associated with acute hemodynamic deterioration, systemic embolization, and high short‑term mortality, particularly when diagnosis or intervention is delayed. Hospitalized patients with mechanical valve thrombosis often present with heterogeneous clinical severity, ranging from subclinical obstruction detected incidentally to cardiogenic shock requiring emergent intervention. 

Identifying reliable prognostic markers at hospital admission is therefore essential for early risk stratification, therapeutic decision‑making, and allocation of critical care resources. However, current clinical practice lacks a simple, validated risk score specifically designed to predict in‑hospital outcomes in this high‑risk population (1).

The pathophysiology of mechanical prosthetic valve thrombosis is complex and multifactorial, involving an interplay between patient‑related risk factors, prosthesis characteristics, anticoagulation status, and transient prothrombotic conditions.

Inadequate anticoagulation remains the most consistently identified risk factor, yet a substantial proportion of patients develop thrombosis despite apparently therapeutic international normalized ratio (INR) values. This observation suggests that factors beyond anticoagulation intensity contribute meaningfully to thrombus formation and its clinical consequences. Comorbidities such as heart failure, advanced age, diabetes mellitus, vascular disease, and prior thromboembolic events may influence not only the likelihood of thrombosis but also the patient’s physiological reserve and ability to tolerate acute valve dysfunction, thereby affecting short‑term mortality during hospitalization (2).

Risk stratification tools have long played a central role in cardiovascular medicine by enabling clinicians to quantify prognosis using readily available clinical variables. Among these, the CHA₂DS₂‑VASc score is one of the most widely adopted scoring systems, originally developed to estimate the risk of stroke and systemic embolism in patients with atrial fibrillation. By incorporating age, sex, heart failure, hypertension, diabetes, prior stroke, and vascular disease, the score reflects cumulative thromboembolic and cardiovascular risk. Beyond its original purpose, CHA₂DS₂‑VASc has increasingly been explored as a broader prognostic marker in diverse cardiovascular conditions, including acute coronary syndromes, heart failure hospitalizations, and valvular heart disease, raising interest in its potential applicability outside atrial fibrillation (3).

The theoretical rationale for evaluating the CHA₂DS₂‑VASc score in patients with mechanical prosthetic valve thrombosis lies in the overlap between the score’s components and known determinants of adverse outcomes in acute cardiovascular illness. Advanced age and heart failure are well‑established predictors of in‑hospital mortality, while diabetes and vascular disease contribute to endothelial dysfunction, systemic inflammation, and impaired microcirculatory reserve. Prior cerebrovascular events may reflect an underlying prothrombotic milieu or advanced atherosclerotic burden. Although female sex is weighted modestly in the score, it may capture sex‑related differences in thrombotic risk and clinical presentation. Collectively, these variables may influence not only thrombus formation but also the severity of clinical presentation and response to treatment in hospitalized patients with prosthetic valve thrombosis (4).

Despite these conceptual considerations, the prognostic role of the CHA₂DS₂‑VASc score in the specific context of mechanical prosthetic valve thrombosis remains insufficiently studied. Existing investigations have primarily focused on predictors of thrombosis occurrence rather than short‑term outcomes such as in‑hospital mortality. Moreover, most studies addressing prosthetic valve thrombosis emphasize echocardiographic findings, valve position, or treatment modality, including thrombolysis versus surgery, with less attention to integrated clinical risk scores. As a result, clinicians often rely on subjective assessment when estimating prognosis, which may lead to variability in management strategies and potentially suboptimal outcomes for high‑risk patients (5).

Hospital mortality in mechanical valve thrombosis is influenced by multiple factors, including the degree of valve obstruction, presence of cardiogenic shock, embolic complications, and delays in diagnosis or intervention. However, baseline patient characteristics play a critical role in determining vulnerability to these acute insults. A simple bedside score that captures cumulative cardiovascular risk could therefore provide incremental prognostic information beyond anatomical and procedural variables. If shown to be associated with in‑hospital mortality, the CHA₂DS₂‑VASc score could serve as an accessible tool to aid early clinical decision‑making, particularly in resource‑limited settings where advanced imaging or invasive hemodynamic monitoring may not be immediately available (6).

Another important consideration is the distinction between predicting thrombotic risk and predicting mortality once thrombosis has occurred. While the CHA₂DS₂‑VASc score was not designed to assess prosthetic valve thrombosis risk per se, its components may be more relevant to outcome prediction after hospitalization. Mortality in this setting often reflects the combined burden of acute valve dysfunction and chronic comorbid disease rather than thrombus burden alone. Thus, even if the score has limited utility in predicting thrombosis development, it may still provide meaningful prognostic insight regarding short‑term survival, length of hospital stay, and need for intensive care or urgent surgical intervention (7).

Evaluating the CHA₂DS₂‑VASc score in hospitalized patients with mechanical prosthetic valve thrombosis also aligns with the broader movement toward repurposing established clinical scores for novel prognostic applications. Such approaches leverage existing clinical familiarity and avoid the complexity of developing entirely new risk models. However, rigorous evaluation is essential to ensure that associations are not merely coincidental and that the score provides independent prognostic value beyond traditional clinical and echocardiographic predictors. Understanding these relationships may also help clarify the mechanisms through which systemic comorbidities influence outcomes in prosthetic valve–related complications (8).

In this context, investigating the prognostic value of the CHA₂DS₂‑VASc score for in‑hospital mortality among patients with mechanical prosthetic valve thrombosis addresses an important gap in the current literature. By focusing on hospitalized patients with confirmed thrombosis, this approach emphasizes clinically meaningful outcomes and real‑world applicability. Clarifying whether a widely used cardiovascular risk score can aid prognostication in this high‑risk population may contribute to improved risk stratification, more informed clinical decision‑making, and ultimately better patient outcomes. Such evidence could also lay the groundwork for future studies integrating clinical risk scores with imaging and laboratory markers to develop more comprehensive prognostic models for mechanical valve thrombosis.

Material and methods

Study design and setting: This was a hospital-based case control study conducted at Shahid Madani Hospital, Tabriz University of Medical Sciences, from March 2017 through March 2022 (corresponding to early 1396 to end of 1400 in the Iranian calendar). All consecutive admissions with a confirmed diagnosis of mechanical prosthetic valve thrombosis (PVT) were screened, and data were captured using a standardized protocol by a trained research team.

Sampling: A census sampling strategy was employed. All eligible patients meeting inclusion criteria during the study window were enrolled until the target sample of 100 individuals was achieved. Cases were defined as hospitalized patients with mechanical PVT; controls were hospitalized patients with mechanical prosthetic valves without PVT during the same period, matched by admission era and ward availability to minimize temporal treatment biases.

Eligibility criteria: Inclusion criteria were: age ≥18 years; presence of at least one mechanical prosthetic heart valve (aortic, mitral, tricuspid, or multivalvular); hospital admission with clinical suspicion of valve dysfunction; and availability of definitive diagnostic confirmation status for PVT by transthoracic or trans esophageal echocardiography and/or fluoroscopy and, when applicable, perioperative findings. Patients were required to have documented anticoagulation data (INR within 48 hours of index evaluation or prior to thrombolysis/surgery), baseline demographics, comorbidities (heart failure, hypertension, diabetes, prior stroke/TIA, vascular disease), and in-hospital outcomes. Exclusion criteria included: bio prosthetic valves only; isolated pannus without thrombus when clearly distinguished by imaging or operative reports; active infective endocarditis with vegetation’s obscuring PVT adjudication; incomplete core variables precluding assignment of CHA₂DS₂‑VASc or outcome assessment; pregnancy; and readmissions of the same event episode (only the first index hospitalization was retained). Patients with sub therapeutic INR on arrival were not automatically excluded; rather, INR status was recorded and adjusted for in analysis to avoid selection bias, while extreme data anomalies (e.g., missing valve position) led to exclusion after blinded adjudication.

Procedures and measurements: Upon enrollment, trained clinicians abstracted data from electronic medical records and bedside charts using prespecified forms. Baseline variables included age, sex, body mass index, cardiovascular history (heart failure, hypertension, diabetes, prior stroke/TIA, vascular disease), smoking status, renal function, and time since valve implantation. Valve‑related information comprised valve position (aortic, mitral, tricuspid), type/model when available, and prior valve interventions. Anticoagulation data encompassed warfarin dose (if applicable), recent INR values (admission and pre-event where documented), use of antiplatelet therapy, and any recent lapses in anticoagulation. The CHA₂DS₂‑VASc score was computed at admission from the standard components using contemporaneous clinical documentation.

Procedures and measurements: Diagnosis of PVT was established by a combination of imaging and clinical criteria: restricted leaflet motion or occlude immobility on cinefluoroscopy; elevated transvalvular gradients or increased Doppler velocities with abnormal effective orifice area on echocardiography; visualization of thrombotic material; and adjudication by the treating cardiologist and cardiac surgeon when surgical findings were available. For controls, the absence of PVT was confirmed by normal or stable prosthetic valve hemodynamics and no clinical/imaging evidence of thrombus during the index hospitalization. Management strategies (anticoagulation optimization, thrombolytic therapy, urgent surgery) and procedural timing were recorded. Complications (systemic embolism, stroke, major bleeding) were prospectively abstracted.

Procedures and measurements: The primary outcome was in-hospital mortality (all-cause) during the index admission. Secondary outcomes included need for intensive care unit admission, vasoactive support, emergent surgical intervention, thrombolysis success/failure, hospital length of stay, stroke, and major bleeding per standard definitions. Data entry underwent double-check verification by independent reviewers, with discrepancies resolved by consensus. To reduce information bias, outcome assessors were blinded to the CHA₂DS₂‑VASc score during adjudication. Continuous variables were inspected for distributional assumptions, outliers were verified against source documents, and a predefined data dictionary ensured uniform variable coding.

Statistical analysis: Continuous variables are presented as mean ± standard deviation or median (interquartile range) as appropriate; categorical variables as counts and percentages. Between-group comparisons (PVT cases vs. controls; survivors vs. non-survivors) used Student’s t test or Mann–Whitney U test for continuous data and chi-square or Fisher’s exact tests for categorical data. The association between CHA₂DS₂‑VASc and in-hospital mortality was examined using multivariable logistic regression, with selection of covariates informed by clinical relevance and univariable screening (p<0.20). Core adjustment covariates included age, sex, valve position, heart failure, diabetes, prior stroke/TIA, vascular disease, renal dysfunction, and admission INR; treatment modality (thrombolysis vs. surgery vs. medical) and presence of shock were evaluated as additional covariates or stratifies in sensitivity analyses. Model calibration and discrimination were assessed using the Hosmer–Lemeshow test and area under the receiver operating characteristic curve (AUC), respectively. We examined linearity of the CHA₂DS₂‑VASc effect with restricted cubic splines and tested for interactions with valve position and anticoagulation status. Robustness was evaluated with: (1) propensity score adjustment for likelihood of receiving thrombolysis; (2) exclusion of patients with missing ancillary covariates; and (3) inverse probability weighting for incomplete INR histories. Two-sided p < 0.05 was considered statistically significant. Analyses were performed using R (version X.Y.Z) or Stata (version XX), as available.

Ethical considerations: The study protocol was approved by the Tabriz University of Medical Sciences Ethics Committee (IR. TBZMED.1400.1256). The research formed Specific Objective No. 3 of thesis number 67370, and was conducted in accordance with the Declaration of Helsinki and relevant national regulations. Given the observational design and use of routinely collected clinical data, the requirement for individual informed consent was waived by the ethics committee; all data were de-identified prior to analysis, and access was restricted to authorized investigators.

Results

Baseline characteristics of patients with mechanical prosthetic valves are presented in Table X. The mean age of the study population was 54.93 ± 11.52 years. Female patients constituted 62% (n=62) of the cohort, while 38% (n=38) were male. Hypertension was present in 31% (n=31) of patients, diabetes mellitus in 22% (n=22), heart failure in 45% (n=45), and a prior cerebrovascular accident in 23% (n=23). Comparison of baseline characteristics between the two study groups revealed no significant differences in age or sex distribution. Similarly, the prevalence of major comorbid conditions including hypertension, diabetes mellitus, heart failure, and prior cerebrovascular accident was comparable between groups, indicating a well‑balanced baseline profile. These findings suggest that underlying demographic and clinical characteristics were unlikely to confound the observed outcome differences between the groups.

The overall mean CHA₂DS₂‑VASc score in the study cohort was 2.58 ± 1.65. When stratified by thrombotic status, the mean score was 2.68 ± 1.63 among patients with thrombosis and 2.23 ± 1.72 among those without thrombosis. Despite a numerically higher average in the thrombosis group, the difference between groups did not reach statistical significance (p=0.23), indicating that the CHA₂DS₂‑VASc score did not discriminate thrombotic risk in this sample (figure 1).

The ROC analysis reveals that the CHA₂DS₂‑VASc score, a widely used clinical tool for assessing thromboembolic risk in atrial fibrillation, demonstrates limited discriminative power when applied to the specific outcome of prosthetic valve thrombosis. With an AUC of 0.417 (95% CI: 0.276-0.558), the score performs only marginally better than a random classifier (AUC=0.5), and the confidence interval straddles the line of no discrimination. The non‑significant p‑value (P > 0.05) confirms that the observed predictive performance does not differ statistically from chance. This finding suggests that the clinical and demographic factors encapsulated by the CHA₂DS₂‑VASc score such as age, sex, congestive heart failure, hypertension, diabetes, prior stroke/TIA, and vascular disease are not sufficiently sensitive or specific to differentiate patients who will develop prosthetic valve thrombosis from those who will not. Consequently, while the CHA₂DS₂‑VASc score remains valuable for guiding anticoagulation in atrial fibrillation, it should not be relied upon as a standalone predictor for thrombotic complications in the context of prosthetic valves. The result underscores the need for alternative risk‑stratification tools or the incorporation of valve‑specific parameters (e.g., valve type, position, flow characteristics, and laboratory markers of hypercoagulability) to improve the early identification and management of this serious complication (figure 2).

In patients without thrombosis, the mean CHA₂DS₂‑VASc score was numerically higher among survivors compared with non‑survivors (2 ± 3 vs. 1 ± 3); however, this difference did not reach statistical significance (P=0.461). This finding suggests that, in the absence of thrombotic events, the CHA₂DS₂‑VASc score was not associated with mortality risk and showed limited discriminatory value for survival outcomes in this subgroup (figure 3). Similarly, among patients with thrombosis, the mean CHA₂DS₂‑VASc scores were comparable between survivors and deceased patients (2 ± 2 vs. 2 ± 3), with no statistically significant difference observed (P=0.127). When comparing across thrombotic status, the lack of significant score separation between alive and deceased individuals in both groups indicates that the CHA₂DS₂‑VASc score does not adequately stratify mortality risk, regardless of the presence or absence of thrombosis. Overall, these results imply that while CHA₂DS₂‑VASc is widely used for thromboembolic risk assessment, it may have limited utility as a prognostic marker for mortality in this clinical context (figure 3).