PURPOSE: Sepsis is a common cause of mortality nationwide. To improve the identification and care of septic patients at our institution, we have provided bundled care, adopting a sepsis bundle based on the Surviving Sepsis Campaign (SSC) guidelines. We hypothesized that we could successfully track the appropriate identification of patients with severe sepsis, as well as their overall mortality, using our hospital discharge database.
METHODS: Our hospital discharge database was prospectively queried over 6 fiscal years, beginning July 1 and ending June 30, annually, during 2005-2010. This period coincided with adoption of the SSC database and care bundles. An annual estimate of the number of severe sepsis cases in our institution throughout this period was made using a method derived from Angus, et al. Physician-diagnosed severe sepsis was identified by ICD-9 codes 995.92, 785.52. Additionally, the database was queried for all-cause mortality of these patients.
RESULTS: In FY 2005, only 171 of 426 patients identified to have severe sepsis by the modified Angus method were given an ICD-9 diagnosis of severe sepsis or septic shock (40%). This grew to 958 out of 1,205 (80%) by FY 2010. Simultaneously, the all-cause in-hospital mortality rate of patients diagnosed with severe sepsis fell from 49% in FY 2005 to 25% in FY 2010. The mortality index (observed mortality/expected mortality) fell from 1.56 in FY 2005 to 0.88 in FY 2010. Average LOS declined from 24.4 days in FY 2005 to 15.9 days in FY 2010. Severe sepsis as a recognized contribution to overall hospital mortality rose from 18.5% in FY 2005 to 48.4% in FY 2010.
CONCLUSIONS: The comparison of the modified Angus method with coded discharges for severe sepsis and septic shock demonstrated improved recognition of severe sepsis. Enhanced recognition, combined with systematic improvements in patient care, effectively reduced the mortality of severe sepsis by approximately 50% over a five-year period.
CLINICAL IMPLICATIONS: The hospital discharge database is an effective tool for measuring the outcomes of performance improvement efforts in severe sepsis.
DISCLOSURE: The following authors have nothing to disclose: Lucas Pitts, Michael Moncure, Chad Cannon, Steven Simpson
No Product/Research Disclosure Information