The importance of clinical prediction models in non-fatal pulmonary embolism: an analysis of the best known clinical scores.

The clinical evaluation in pulmonary embolism (PE) is the first instrument used by practitioners in the management of this potentially fatal pathology. The necessity of develop- ing certain valid and especially affordable practical instruments has led to the emergence of various clinical prediction models. The purpose of this paper is to analyze the main clinical scores, as a diagnostic or a prognostic tool, with their strengths and weaknesses. The PESI score, while relatively recent, remains the most investigated and validated prognostic score for the identification of the mortality risk and major adverse events, with economic implications of health services reduction costs through the accurate identification of patients with a low risk who are candidates of early hospital discharge. The simplified Geneva score (with a similar accuracy as the Geneva one) identifies a high or low PE probability, especially in combination with D-dimers, with a prognosis value as well. The Wells and simplified Wells scores identify the high or low probability, being improved by the level of D-dimers, having similar results with the Geneva score. The LR-PED score, conceived as an identification score for low risk, uses biochemical and electrocardiographic markers, but is less validated. The Vienna Prediction Model is another system for the evaluation of the recurrence in which the level of D-dimers is the main prediction factor. Other scores were evaluated with a statistically low significance. The Geneva and the PESI scores remain the most valuable instruments of diagnosis and clinical prognostic, respectively.