Inverse relationship of serum albumin to the risk of venous thromboembolism among acutely ill hospitalized patients: Analysis from the APEX trial

Abstract

Hypoalbuminemia is a common finding and independent predictor for unfavorable prognosis. The prognostic value of albumin measurement for short-term VTE prediction in hospitalized patients remains unclear. In the APEX trial (ClinicalTrials.gov identifier: NCT01583218), medical inpatients were randomized to receive either extended-duration betrixaban or shorter-duration enoxaparin and followed for 77 days. Baseline albumin concentrations were obtained in 7266 subjects with evaluable VTE endpoints. The association of baseline albumin to VTE was assessed, with adjustment for patient characteristics, thromboprophylaxis, and biomarkers for fibrinolysis and inflammation (ie, D-dimer and C-reactive protein [CRP]). VTE risk refinement was evaluated by incorporation of albumin to well-validated risk assessment models. A stepwise increase in the risk of VTE (P < .0001) was observed with lower levels of albumin. Patients at the bottom albumin quartile (<35 g/L) had a two-fold greater odds for developing VTE compared with the top quartile (≥42 g/L) (OR = 2.119 [95% CI, 1.592-2.820]; adjusted OR = 2.079 [1.485-2.911]). The odds for VTE increased by 1.368 (95% CI, 1.240-1.509) times per SD decrement of albumin (5.24 g/L). Compared with the propensity score-matched pairs of patients with albumin ≥35 g/L, patients with albumin <35 g/L had a greater risk of VTE (OR = 1.623 [1.260-2.090]; adjusted OR = 1.658 [1.209-2.272]). Albumin measurement also refined VTE risk discrimination and reclassification after inclusion in the risk assessment models. In conclusion, acutely ill hospitalized patients with low serum albumin had an increased VTE risk through 77 days. VTE risk assessment models for medical inpatients should consider incorporation of baseline albumin measurement.