PURPOSE: Hospital discharge databases are a valuable resource for epidemiologic studies of severe sepsis (SS). Our aim is to compare the occurrence, resource utilization, characteristics, and outcome of patients with severe sepsis identified by two published definitions.
METHODS: We conducted a retrospective study based on hospital discharge and charge master information of ∼5 million discharges from ∼500 US hospitals during 2006 (Premier PerspectiveTM Database). SS was defined via ICD-9 coding following definitions used in Angus et al, 2001 (Crit Care Med 29:1303–1310) (A definition) and Dombrovskiy et al, 2007 (Crit Care Med 35:763–768) (D definition). Both definitions required infection and an acute organ dysfunction from one of six systems: respiratory, cardiovascular(CV), coagulation, renal, hepatic, and central nervous system. A major difference is that a septicemia code was required for the D definition, while in the A definition, all bacterial or fungal infections were included.
RESULTS: Incidence rates of SS per discharge were 6% and 1.7% using A and D definitions, respectively. SS patients identified with A vs. D definitions had the following characteristics: patients with two or more organ dysfunction: 29.4% vs. 52.3%; with either respiratory, CV or both organ dysfunctions: 61.3% vs. 73.6%; in-hospital mortality: 15.6% vs. 31.1%; mean length of hospital stay (day): 11.4 vs. 14.9, and mean hospital costs: $23,000 vs. $33,500. However, the Charlson's comorbidity index was similar in these two groups (mean): 2.70 vs. 2.77.
CONCLUSION: Severe sepsis patients identified using Dombrovskiy's definition appears to represent a set of patients at higher risk of death than those patients identified using Angus's definition. Blood stream infection (septicemia coding), but not comorbidity, may be associated with the increased risk. Further evaluation is warranted to understand which definition truly reflects the severe sepsis population.
CLINICAL IMPLICATIONS: ICD-9 codes can be used to retrospectively capture patients with a severe sepsis. The specific ICD9 codes used to define severe sepsis may affect research and healthcare decision making in severe sepsis.
DISCLOSURE: Huiying Yang, Shareholder Shareholder in AstraZeneca; Employee Employee of AstraZeneca; No Product/Research Disclosure Information