A way to select on clinical grounds patients with high risk for pulmonary embolism: a retrospective analysis in a nested case-control study