Clinical scoring systems in meningococcal disease

Scoring systems used in meningococcal disease have been developed and validated to predict death or severity of illness in cohorts of patients, usually in the setting of intensive care. When using these scores, it is important to remember that any individual's score is only of limited value in prognosticating about that individual patient and can be misleading (1). The accuracy of any score is determined by how well its estimated pattern of mortality compares with that observed in the score's developmental cohort, rather than in individual patients. Scores cannot be considered interchangeable; different scores, comprised of different sets of variables to arrive at the probability of mortality, may arrive at the same overall aggregate risk of mortality in a population of patients with meningococcal disease, despite individual estimates within this cohort of patients differing widely between scores. The ability of a score to predict death for patients who die and survival for those that live is best-described by the area under the receiver operator characteristic (ROC) curve (2). This curve is a plot of sensitivity versus "(1 -specificity)", i.e., the true-positive to false-positive fractions, at different decision thresholds. The greater the discriminative ability of a test, the closer the area under the curve comes towards 1.0, where the true-positive fraction is 1.0 or 100% (perfect specificity) and the false-positive fraction is 0 (perfect sensitivity). No score, however, has sufficient accuracy for individual prognostication and the use of scores in this way is inappropriate for clinical practice or research (3).