[Clinical prediction rules: a reader's guide]

Abstract

Introduction: Clinical prediction rules or scores are instruments that allow to quantify the respective contributions of medical history, clinical examination and laboratory tests to a diagnosis or a prognosis in a given patient.

State of the art: The development of clinical prediction rules requires three steps: 1) derivation of the rule from a population with known clinical characteristics and diagnosis or prognosis; 2) validation of the rule in a population distinct from the derivation sample; and 3) evaluation of the impact of the rule on clinician behavior and patient outcomes.

Perspectives: This paper provides a guide for the critical appraisal of a prediction rule illustrated by two clinical scores for the assessment of the clinical probability of pulmonary embolism.

Conclusions: Clinical prediction rules or scores have become indispensable components of the clinician's toolbox.