Peri-Diagnostic D-Dimer-to-Albumin Ratio and Intensive Care Unit Admission in Pulmonary Embolism: A Single-Center Retrospective Cohort Study

Background/Objectives: Characterizing factors associated with intensive care unit (ICU) admission among hospitalized patients with pulmonary embolism (PE) is clinically relevant, particularly when established risk-stratification variables are incompletely documented in routine electronic medical records. The D-dimer-to-albumin ratio (DAR) combines a marker of fibrin turnover with serum albumin, a marker of systemic vulnerability; however, whether DAR provides clinically meaningful information beyond D-dimer alone remains uncertain. We evaluated the association between peri-diagnostic DAR and ICU admission and compared DAR with its component biomarkers. Methods: This single-center retrospective cohort study used electronic medical record data from 2009 to 2024. Candidate PE-related hospitalizations were identified from diagnostic records and clinically reviewed to verify a current acute PE event. Patients were eligible when DAR were measured within ±48 h of the PE diagnosis timestamp in the electronic medical record; the first eligible hospitalization per patient was retained. The primary outcome was ICU admission. Logistic regression models were adjusted for age, sex, recorded malignancy, and concomitant deep vein thrombosis (DVT). Results: Among 159 clinically reviewed patients with current acute PE, 143 (89.9%) had documented objective imaging confirmation, and 22 (13.8%) were admitted to the ICU. Each 1-unit increase in log-DAR was associated with greater odds of ICU admission (odds ratio [OR], 2.02; 95% confidence interval [CI], 1.43–2.86; p < 0.001), and the highest DAR quartile (Q4) was associated with greater odds than Q1–Q3 (OR, 6.78; 95% CI, 2.52–18.20; p < 0.001). However, in a temporally restricted analysis including 145 patients and 8 subsequent ICU admissions, the associations were attenuated and no longer statistically significant. Log-DAR was almost perfectly correlated with log-D-dimer (Spearman ρ = 0.995) and did not materially improve discrimination (area under the receiver operating characteristic curve [AUC], 0.788 vs. 0.785; ΔAUC, 0.003; p = 0.471; optimism-corrected AUC, 0.738 vs. 0.737). Conclusions: Peri-diagnostic DAR was associated with ICU admission in the primary analysis, but the temporally restricted analysis did not support prospective prediction of subsequent ICU admission. Its incremental value beyond D-dimer appeared limited, and DAR should be considered an exploratory adjunct rather than a replacement for established PE risk-stratification tools.