Models for Assessing Severity of Illness in Patients with Bloodstream Infection: a Narrative Review

To identify the severity of illness (SOI) models that have been utilized most often in studies of bloodstream infection (BSI) and analyze their development to determine the model that is most useful for measuring SOI. In 555 of 945 studies of BSI that utilized a SOI model, the Pitt bacteremia score (PBS) or APACHE model was used in 91% of the studies. Limitations of the PBS and APACHE model were identified in an analysis of the development of these models. There has also been concern about measuring SOI at the time of first identifying BSI because measurements at that time reflect the effects of infection making it an intermediate variable rather than a confounding variable. It has been recommended that SOI be measured prior to identification of BSI. Neither the PBS nor APACHE model can be considered the best method for measuring SOI in BSI studies. An important consideration is the timing of measurement of SOI but the optimum time for this prior to the first positive blood culture has not been determined. Future studies of BSI need to focus on identifying the best model for measuring SOI as well as the optimum time for applying the model.