OBJECTIVE: The objective of this study was to utilize hospital EMR data to study severe sepsis, including the timing of organ dysfunction and source of infection. These study variables are needed to fully investigate clinical and economic outcomes in severe sepsis. METHODS: Inpatient, pharmacy, laboratory, and UB-92 billing data from Cerner's Health Facts® database January 1, 2001 – September 30, 2007 were used. Qualifying patients had an inpatient encounter with a primary or secondary diagnosis of sepsis (ICD-9-CM code: 038.x, 040.82, 785.52, 995.91, 995.92) AND evidence of ≥1 acute organ system dysfunction within 24 hours prior to index blood culture through 3 days post admission AND an order for a blood culture and IV antibiotic within ±24 hours of admission. Organ dysfunction was defined to be consistent with a Sequential Organ Failure Assessment score of ≥2 for the following categories: respiratory, hematologic, hepatic, cardiovascular, and renal. Sepsis source was identified by ICD-9. RESULTS: 11,761 patients met the study definition of severe sepsis. The mean age was 69 years and 43% were 75 years old or older. Cardiovascular (51%), renal (51%) and respiratory (49%) dysfunctions were the most common organ dysfunctions in the study cohort. Approximately 70% had 1 or 2 organ dysfunctions; 26% had three or four organ dysfunctions; only 3% had all 5 organ dysfunctions. The sequence of organ dysfunction (median time from index blood culture) was as follows: respiratory < cardiovascular < renal < hepatic < hematologic. Nearly half of the patients had either multiple or unknown sources of infection followed by respiratory (20%), genitourinary (14%), gastrointestinal (8%), and skin/soft tissue (6%). CONCLUSION: Hospital EMR data can be used to study severe sepsis, including patterns of organ dysfunction and source of infection. Additional research can quantify burden-of-illness by source of sepsis and lead to earlier recognition of developing sepsis.