Project Summary

The most fundamental aspect of all research and clinical knowledge about sepsis is its definition. But sepsis is a syndrome rather than a disease, and clinicians and epidemiologists struggle to define and identify sepsis. Currently, two approaches exist for sepsis case-finding in epidemiologic work. Both seek to identify patients with infection and acute organ failure. The first uses diagnostic codes in administrative hospital records. There are a variety of these algorithms, giving not only different incidence rates of sepsis, but also differences over time. The second approach uses data from electronic medical records. Not only do these approaches give different sepsis incidence rates, but there is evidence suggesting that both underestimate the true incidence of sepsis. In this project we will look to refine/improve both existing methods, and to develop a profoundly different approach to identifying sepsis; this will use advanced statistical methods to large, population-based patient datasets, seeking self-defined clusters that can be interpreted as representing phenotypes along the clinical spectrum of sepsis.  

If you would like to collaborate on this research project or learn more about it, please contact the Principal Investigators.