Blood Urea Nitrogen and Atrial Fibrillation Identify Geriatric Peri-Hospitalization Cardiovascular and Cerebrovascular Vulnerability After Hip Arthroplasty

Technical Summary: Blood Urea Nitrogen and Atrial Fibrillation Identify Geriatric Peri-Hospitalization Cardiovascular and Cerebrovascular Vulnerability After Hip Arthroplasty

Problem Statement

Elderly patients undergoing hip arthroplasty face significant risks of in-hospital Major Adverse Cardiovascular and Cerebrovascular Events (MACCE), which lead to prolonged hospitalization, delayed functional recovery, and increased mortality. Current perioperative risk assessment tools, such as the Revised Cardiac Risk Index (RCRI), ASA classification, and the Nottingham Hip Fracture Score, were largely developed for broad non-cardiac surgical cohorts or focused on all-cause mortality rather than specific in-hospital MACCE. These conventional tools often lack validation in the specific context of geriatric hip surgery and fail to incorporate readily available, routine laboratory markers—such as Blood Urea Nitrogen (BUN)—that may serve as sensitive proxies for physiological reserve and renal-cardiovascular stress. Consequently, there is a gap in identifying a specific, interpretable risk phenotype for in-hospital MACCE in older adults that integrates routine clinical and laboratory variables.

Methodology

This retrospective cohort study utilized the MIMIC-IV version 3.1 database, a deidentified electronic health record dataset from Beth Israel Deaconess Medical Center.

• Cohort: The study included 1,253 patients aged 65 years or older who underwent hip arthroplasty (total, partial/hemiarthroplasty, or revision/spacer-related).
• Outcomes:
  • Primary Endpoint: Broad-definition in-hospital MACCE, defined as a composite of acute myocardial infarction, ischemic/hemorrhagic stroke, cardiac arrest, cardiogenic shock, acute heart failure, or in-hospital death.
  • Sensitivity Analysis: A "strict" MACCE endpoint was used, excluding cardiogenic shock and acute heart failure to test endpoint specificity.
• Predictors: The analysis incorporated demographic data, admission acuity (urgent/emergency vs. elective), comorbidities (coronary artery disease, atrial fibrillation, diabetes, chronic kidney disease), and routine inpatient laboratory variables (hemoglobin, creatinine, BUN, glucose, etc.) captured from admission up to the surgery date.
• Statistical Approach:
  • Multivariable logistic regression was used to identify independent risk factors.
  • A parsimonious "clinical core model" was constructed using five variables: age, urgent/emergency admission, coronary artery disease, atrial fibrillation, and peak admission BUN.
  • Model performance was evaluated using repeated stratified 5-fold cross-validation, assessing discrimination (AUC, AUPRC), calibration, and Brier scores.
  • Sensitivity analyses included strict endpoint definitions and a conservative preprocedure laboratory window (excluding values on the surgery date).
  • Subgroup analyses were performed based on hip fracture diagnosis and admission acuity.
  • A graphical nomogram was generated to visualize risk contributions.

Key Contributions

1. Identification of a Renal-Cardiovascular Vulnerability Profile: The study identifies a distinct risk phenotype characterized by the convergence of elevated BUN and atrial fibrillation, suggesting a synergistic effect between renal-metabolic stress and cardiac rhythm instability in geriatric hip surgery patients.
2. Development of a Parsimonious Clinical Core Model: The authors propose a five-variable model (Age, Urgent/Emergency Admission, Coronary Artery Disease, Atrial Fibrillation, and BUN) that captures the majority of the risk stratification signal for in-hospital MACCE without requiring complex calculations or unavailable data.
3. Integration of Routine Biomarkers: The study demonstrates that BUN, a standard and inexpensive laboratory test, serves as a robust predictor of MACCE, potentially outperforming or complementing traditional risk scores that often overlook renal-metabolic stress markers.
4. Visual Risk Summary: A nomogram was created to translate the statistical model into an intuitive, points-based visual tool for bedside reference, illustrating the stepwise increase in risk associated with BUN strata and the accumulation of core risk factors.

Results

• Incidence: Broad MACCE occurred in 81 patients (6.46%), and strict MACCE occurred in 44 patients (3.51%). Acute heart failure was the most frequent component of the broad endpoint.
• Independent Risk Factors: Older age, urgent/emergency admission, coronary artery disease, atrial fibrillation, and higher BUN were independently associated with broad MACCE.
• Dominant Signals: BUN and atrial fibrillation showed the most consistent predictive signals across primary, strict-endpoint, and preprocedure-window analyses.
  • BUN: Higher BUN levels showed a strong association with MACCE (OR 1.68 per 10 mg/dL). Risk increased stepwise with BUN levels: 3.6% for BUN <20 mg/dL, rising to 24.8% for BUN >30 mg/dL.
  • Atrial Fibrillation: Patients with atrial fibrillation had significantly higher MACCE rates (OR 2.88 for broad MACCE).
• Risk Gradients:
  • Patients with four or more of the five core risk factors had a MACCE rate of 31.4%, compared to 4.0% for those with 0–1 factors.
  • The clinical core model performed comparably to the full 12-variable model and supplementary tree-based algorithms (Random Forest, XGBoost).
• Subgroup Consistency: The predictive value of elevated BUN and atrial fibrillation remained consistent across both hip fracture and non-fracture subgroups, as well as across urgent/emergency and elective admission contexts.

Significance and Claims

The paper claims to provide a "geriatric peri-hospitalization vulnerability framework" that complements existing risk scores. By highlighting elevated BUN and atrial fibrillation as dominant signals, the study suggests that orthogeriatric teams can implement tiered admission triage. Specifically, patients presenting with this dual risk phenotype (elevated BUN + atrial fibrillation) may benefit from:

• Early cardiac consultation.
• Targeted volume optimization.
• Continuous rhythm monitoring.
• Enhanced postoperative cardiac telemetry and daily volume monitoring.

The authors emphasize that this five-factor framework is designed as a complementary adjunct, not a replacement, for established guidelines like the RCRI. It aims to add orthogeriatric-specific granularity to standard workflows by utilizing universally available, low-cost variables to identify patients with acute physiological stress who might be misclassified as low-risk by traditional scores.

Limitations Acknowledged:
The authors note that the study is a single-center retrospective analysis with non-random missingness in laboratory data (selection bias). The dataset lacks formal measures of frailty, cognition, and nutritional status. Furthermore, the model has not undergone external multicenter validation, and the strict MACCE event count was relatively low. Therefore, the nomogram is presented as an illustrative risk summary rather than a validated clinical decision-making tool, and prospective validation is required before integration into clinical guidelines.