| Literature DB >> 9741250 |
B S Dahler-Eriksen1, I Brandslund, J F Lassen, T Lauritzen.
Abstract
Conflicting data for predictive values for C-reactive protein (CRP) in its ability to distinguish between viral and bacterial diseases are reviewed. Study designs regarding setting, patient-mix, severity of disease and prevalence seem to determine the magnitude of predictive values. We have calculated predictive values for patients suspected of septicaemia, meningitis, appendicitis, cholecystitis, upper- and lower respiratory disease, acute sinusitis and acute otitis media, and revealed the highest predictive values among patients suspected for severe and generalized infections. More localized diseases have lower predictive values. We emphasize the importance of a study design where the circumstances resemble the real situations in which the test is supposed to be used. This will ensure the clinical applicability of predictive values for a diagnostic test.Entities:
Mesh:
Substances:
Year: 1998 PMID: 9741250
Source DB: PubMed Journal: Ugeskr Laeger ISSN: 0041-5782