New Mathematical Model of Sepsis Developed To Predict Outcomes

A new mathematical model of sepsis can help predict deaths, discharges, and disease progression in hospital patients with this serious bacterial blood infection.

Investigators from the University of Pittsburgh (PA, USA) used data from a large, multi-center study to develop a dynamic micro-simulation model that simulates changing health over time, as represented by the Sepsis-Related Organ Failure Assessment (SOFA) score, as a function of a patient's previous health state and length of hospital stay.

The researchers used data from patients enrolled in the GenIMS (Genetic and Inflammatory Markers of Sepsis) study to calibrate the model, and tested the model's ability to predict deaths, discharges, and daily SOFA scores over time using different algorithms to estimate the natural history of sepsis. This included information on admission date, movement between wards, trips to the intensive care unit (ICU), discharge, deaths, and disease progression from more than 1,800 patients with pneumonia-related sepsis.

The researchers found that the model closely predicts changing health and the pattern and number of discharges and deaths in patients over a 30-day period. There were 1,776 discharges in the original multi-center study, and based on the precision of its patient-matching algorithms, the model predicted between 1,779 and 1,804. The model forecast between 62 and 84 of the 85 patients who actually died. The researchers also found the simulation model could predict not only the number but also the pattern of events over time, although the ability to predict when deaths and discharges occur over time varies. The new sepsis model was described on June 14, 2007, in the online open access journal Critical Care.

"The model is able to predict hospital discharges, in-hospital deaths, and serial SOFA scores of patients with sepsis, and it supports the assertion that the duration of disease is a critical factor in predicting the outcomes of sepsis,” concluded lead author Görkem Saka, M.S., and colleagues.


Related Links:
University of Pittsburgh