Level 1: Likely reliable evidence
For the clinical prediction of probability of pulmonary embolism (PE), the Pisa model is effective, based on a derivation cohort study of 1,100 patients with suspected PE (40% had PE) and a validation cohort study of 400 patients with suspected PE (41% had PE) (Am J Respir Crit Care Med. 2008;178:290-294). PE was confirmed by angiography or autopsy and ruled out by angiography, autopsy, or normal perfusion scan. Patients with a normal perfusion scan were followed up at six months.
The model includes 10 variables positively associated with PE and six variables negatively associated with PE. Positive variables are older age (57-67 years, 68-74 years, 75 years and older), male gender, immobilization, history of deep venous thrombosis, sudden onset of dyspnea, chest pain, fainting or syncope, hemoptysis, unilateral leg swelling, and ECG with acute cor pulmonale. Negative variables are history of cardiovascular disease, history of pulmonary disease, orthopnea, fever >38°C (100.4°F), wheezes, and crackles.
In the derivation cohort, the prevalence of PE was 4% when predicted clinical probability was slight (0%-10%), 26% when moderate risk was predicted (11%-50%), 65% when substantial risk was predicted (51%-80%), and 91% when high risk was predicted (80%-100%). In the validation cohort, the prevalence of PE was 2% when predicted clinical probability was slight, 28% when moderate risk was predicted, 67% when substantial risk was predicted, and 94% when high risk was predicted.
Two calculators based on the Pisa model are available online. One calculator model does not need chest x-ray findings (medcalc3000.com/PulmonaryEmbRiskPisa.htm or
www.ifc.cnr.it/pisamodel/pisamodel2/calcolo2.html, both accessed September 16, 2008) and the other model uses chest x-ray findings (medcalc3000.com/PulmonaryEmbRiskPisaCXR.htm or www.ifc.cnr.it/pisamodel/pisamodel1/calcolo.html, both accessed September 16, 2008).