PURPOSE: To describe the epidemiology of hospitalization of elderly Americans for septicemia or bacteremia. METHODS: Medicare claims data for discharges from 1991 through 1998 were used to study 75,920 hospitalizations with the principal diagnosis of septicemia or bacteremia in patients aged 65 years or older. RESULTS: "Unspecified septicemia" was the commonest principal diagnosis, followed by septicemia due to Escherichia coli or staphylococci. From 1991 through 1997, annual discharges for "unspecified septicemia" increased 108%, and those for pneumococcal septicemia increased 310%. Decreases in reported septicemia were seen after increases in the proportion of beneficiaries in Medicare health maintenance organizations. Discharge rates for septicemia principal diagnoses increased steeply with age. Age-specific discharge rates were usually highest fur black men and lowest for white women. Exceptions included septicemia due to E. coli, with white men at low risk, and pneumococcal septicemia, without significant differences between races or sexes. The case-fatality rate in hospital ranged from 4.2% with "bacteremia" and 6.9% with E. coli septicemia to 22.2% with "septicemia due to gram negative organism, unspecified," and 26.8% with "unspecified septicemia." Staphylococcal septicemia, septicemia due to pseudomonas, and septicemia due to anaerobes were the costliest common principal diagnoses in terms of the mean duration of hospital stay. CONCLUSIONS: Unexplained sharp increases were reported in hospitalization for septicemia or bacteremia in elderly Americans. Marked variation by race and sex were evident in discharge rates with these principal diagnoses. Prognosis and average cost of treatment also differed substantially among common rubrics. Further investigation of individual diagnoses should concentrate on explaining secular trends, exploring the basis for variation by race and sex, and elucidating risk factors for poor clinical outcomes. Ann Epidemiol 2001;11:118-126. (C) 2001 Elsevier Science Inc. All rights reserved.
