International Journal of Antimicrobial Agents 30S (2007) S88–S92 Risk scoring and bloodstream infections Evelina Tacconelli a,b,∗ , Maria A. Cataldo a , Giulia De Angelis a , Roberto Cauda a a Istituto Malattie Infettive, Universit` a Cattolica S. Cuore, Rome, Italy b Division Infectious Diseases, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, USA Abstract Risk-scoring systems are utilised in patients with bloodstream infections (BSI) to quantify disease-associated morbidity and mortality based on simple clinical or laboratory data usually obtained early in the course of illness. In order to reduce BSI-associated mortality, specific scores were elaborated to allow early diagnosis and prompt and appropriate antibiotic therapy. Risk scoring was also successfully derived and validated to identify patients at higher risk for antibiotic-resistant BSI, or colonisation, mainly due to methicillin-resistant Staphylococcus aureus and vancomycin-resistant enterococci. However, a major limitation of risk-scoring systems is the relevance to the local epidemiological environment and the difficulty in generalising results from a single study. Intelligence technology recently utilised scores to predict risks for specific pathogens causing BSI. An example of this innovation, the TREAT system, was able to significantly reduce mortality, length of hospitalisation and costs in patients with BSI. New randomised clinical trials are needed to study the efficacy of clinical scores in reducing BSI-associated morbidity and mortality. © 2007 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. Keywords: Bloodstream infection; Risk scoring; Mortality; Antibiotic resistance; Diagnosis; Therapy 1. Introduction Predicting the risk of a specific disease has assumed an added importance in the modern era of evidence-based medicine. Risk-scoring systems are utilised, in many fields of medicine, to quantify disease-associated morbidity and mor- tality based on clinical or laboratory data usually obtained early in the course of illness. Validated scoring systems are useful in the risk stratification of patients, comparison of var- ious therapeutic options, decision making regarding intensity of patient monitoring, and determination of unexpected mor- tality. In high-risk populations such as those admitted to the intensive care unit (ICU), several scores, such as the child classification for liver failure [1], the Acute Physiology And Chronic Health Evaluation (APACHE) [2] and the Sequential Organ Failure Assessment (SOFA), are already widely used to predict outcomes [3]. An ideal risk scoring should be user-friendly, based on commonly available variables, cost-effective, and should be ∗ Corresponding author. Tel.: +39 06 30154945; fax: +39 06 3054519. E-mail address: etaccone@bidmc.harvard.edu (E. Tacconelli). capable of being readily incorporated into pre-existing audit programmes. When using a risk scoring, the method of its derivation and validation should be understood to understand whether the population of patients used to develop the mod- els is related to another patient population. A risk scoring generated for a group of patients may be useful for the popu- lation of other wards or hospitals only if an adjustment for the patient’s risk is made. Moreover, models may become obso- lete if they do not reflect current management and treatments. Risk scoring may be helpful in general practice once it can demonstrate accuracy and effectiveness in improving health outcomes. In the infectious disease area and most of all in the management of bloodstream infections (BSI) and pneumo- nia, risk scoring is widely and effectively used mainly to assess severity of illness and risk of mortality. The main difficulty in elaborating risk scoring for BSI is related to the complexity and heterogeneity of the disease. The clini- cal presentation, the risk of metastatic infections and other complications, the rate of cure with antimicrobial ther- apy and the mortality rate, differ widely depending on the aetiologic agent and on antibiotic susceptibility pattern, 0924-8579/$ – see front matter © 2007 Elsevier B.V. and the International Society of Chemotherapy. All rights reserved. doi:10.1016/j.ijantimicag.2007.06.013