Sensitive, specific, predictive… statistical basics: how to use biomarkers

Abstract

Biomarkers are frequently used in critically ill patients, especially during inflammatory and/or infectious diseases such as severe sepsis and septic shock. The rationale of when to measure laboratory parameters, which marker may be useful, and how to interpret the results is not well defined. Terms like sensitive, predictive, or significant to describe the capabilities of specific markers are often mixed up or misused, which may have fatal consequences regarding diagnosis and treatment. This review reflects some statistical basics with clinical examples, showing possibilities as well as limitations of how data for biomarkers may be used in critically ill patients.