The significant negative impact of in-hospital venous thromboembolism after cardiovascular procedures

Abstract

Our objective was to assess the impact of venous thromboembolism (VTE) on common postoperative cardiovascular surgical patients. An administrative database, the Nationwide Inpatient Sample (NIS, a sampling of 20% of all inpatients across the United States), from 1998 to 2001, was queried for all patients who were hospitalized for primary procedural diagnosis of abdominal aortic aneurysm repair (AAA), amputation (AMP), coronary artery bypass grafting (CABG), carotid endarterectomy (CEA), lower extremity revascularization (LE), and aortic or mitral valve repair (VALV) and for secondary diagnosis of VTE, using standard ICD-9-CM codes. To validate these findings for specificity, the same patient procedure groups with a secondary code of VTE and the same hospitalization procedural code for inferior vena cava (IVC) filter were also analyzed. Factors relating to VTE and the outcomes of death, length of stay (LOS), and unfavorable discharge were analyzed by logistical regression, with odds ratios (ORs) reported as well as analysis of covariance for cost and LOS determinations. A total of 191,666 patients were identified from the NIS, with a mean age of 68 years, 65% men, 85% white race, and a mean VTE incidence of 0.68%. VTE incidence varied with primary procedure: AAA = 1.2%, AMP = 1.1%, CABG = 0.54%, CEA = 0.26%, LE = 0.78%, VALV = 0.63%. VTE was more likely with AAA (OR = 3.9), AMP (3.1), LE (2.8), VALV (2.0), and CABG (1.9) (all compared with CEA, P < 0.0001) and female gender (1.14, P =0.03) but not race or age. Mortality was associated with increased age (1.05), female gender (1.2), and VTE (3.4) (all P < 0.0001). Inpatient costs were 14% higher (P < 0.001), and LOS was increased by 68% compared with those patients not having a VTE (P < 0.0001). Unfavorable discharge status was associated with increased age (1.05), female gender (1.4), and VTE (2.2), among others. A similar magnitude of effect was observed with the subgroup analysis (n = 150) of those undergoing the index procedures who received an IVC filter during the same hospitalization as VTE diagnosis, including increased risk of death, cost, LOS, and unfavorable discharge (all P < 0.001). The occurrence of VTE is not common and varies with the cardiovascular procedure but significantly increases the risk of in-hospital death, cost, LOS, and unfavorable discharge. Process variables should be examined to identify practice patterns that might better prevent this complication.