Objective: To evaluate the possible benefit of inserting a caval filter for the prevention of pulmonary embolism (PE) in an unselected group of patients admitted to hospital.
Design: Analysis of clinical data collected prospectively.
Setting: University hospital, Sweden.
Subjects: 1391 patients who presented during 1987 435 with deep vein thrombosis (DVT), thrombus in the right atrium or the right ventricle or PE at necropsy; 366 patients shown to have DVT on phlebography; 44 shown to have PE on pulmonary scintigraphy; and 546 patients operated on for hip fractures.
Outcome measures: Clinical description of patients and groups of patients who died of PE who could possibly have benefitted from insertion of a filter.
Results: Of the 435 patients with DVT or PE or both, 141 had DVT in the femoral or iliac veins or in the inferior vena cava together with PE which contributed to the cause of death. Only 11 of these had been suspected clinically of having DVT in or below the inferior vena cava and only 1 of these was less than 80 years old and did not have malignant disease with distant metastases. Of the 366 patients with phlebographically confirmed DVT 8 died with PE that had contributed to the cause of death; 6 had incurable cancer and the remaining two were over 85 years. Of the 44 patients with scintigraphically confirmed PE 2 patients (aged 78 and 89) died of PE that had already been diagnosed clinically and none would have benefited from insertion of a filter after diagnosis. There were 9 PE deaths related to PE among 546 patients operated on for hip fractures, and all but one had potentially lethal coexisting disease or were over 80 years old.
Conclusion: In this epidemiological survey only a few patients would have benefited from insertion of a caval filter. The results call for a restricted use of caval filters until benefit has been confirmed by prospective studies.