Clinical prediction rules in primary care: what can be done to maximise their implementation?

Clinical prediction rules (CPRs) have become more prevalent in the published literature in recent years. Known by an array of synonymous terms including risk score, scorecard, algorithm, guide, and model, CPRs are clinical tools that quantify the contribution of a patient’s history, physical examination, and diagnostic tests to stratify patients in terms of the probability of having a specific target disorder. Outcomes of CPRs can be presented as diagnosis, prognosis, referral, or treatment. Although not designed to replace clinical knowledge and experience, CPRs do offer a way to assist with the overall diagnostic and prognostic process.[1] Despite the value of these clinical tools, relatively few CPRs have been quantified and their utility validated. One CPR that has gained widespread acceptance is the Centor score,[2] which is based on four clinical features (tonsillar exudate, tender cervical anterior adenopathy, history of fever, and absence of cough) and is used to identify patients with group A beta-haemolytic streptococcal throat infections. What can be done to expedite implementation of other CPRs into routine primary care?