Health care--associated bloodstream infections in adults: a reason to change the accepted definition of community-acquired infections.

BACKGROUND: Bloodstream infections occurring in persons residing in the community, regardless of whether those persons have been receiving health care in an outpatient facility, have traditionally been categorized as community-acquired infections.
OBJECTIVE: To develop a new classification scheme for bloodstream infections that distinguishes among community-acquired, health care-associated, and nosocomial infections.
DESIGN: Prospective observational study.
SETTING: One academic medical center and two community hospitals. PATIENTS: All adult patients admitted to the hospital with bloodstream infection. MEASUREMENTS: Demographic characteristics, living arrangements before hospitalization, comorbid medical conditions, factors predisposing to bloodstream infection, date of hospitalization, dates and number of positive blood cultures, results of microbiological susceptibility testing, dates of hospital discharge or death, and mortality rates at 3 to 6 months of follow-up.
RESULTS: 504 patients with bloodstream infections were enrolled; 143 (28%) had community-acquired bloodstream infections, 186 (37%) had health care-associated bloodstream infections, and 175 (35%) had nosocomial bloodstream infections. Of the 186 patients with health care-associated bloodstream infection, 29 resided in a nursing home, 64 were receiving home health care, 78 were receiving intravenous or intravascular therapy at home or in a clinic, and 117 had been hospitalized in the 90 days before their bloodstream infection. Cancer was more common in patients with health care-associated or nosocomial bloodstream infection than in patients with community-acquired bloodstream infection. Intravascular devices were the most common source of health care-associated and nosocomial infections, and Staphylococcus aureus was the most frequent pathogen in these types of infections. Methicillin-resistant S. aureus occurred with similar frequency in the groups with health care-associated infection (52%) and nosocomial infection (61%) but was uncommon in the group with community-acquired bloodstream infection (14%) (P = 0.001). Mortality rate at follow-up was greater in patients with health care-associated infection (29% versus 16%; P = 0.019) or nosocomial infection (37% versus 16%; P < 0.001) than in patients with community-acquired infection.
CONCLUSIONS: Health care-associated bloodstream infections are similar to nosocomial infections in terms of frequency of various comorbid conditions, source of infection, pathogens and their susceptibility patterns, and mortality rate at follow-up. A separate category for health care-associated bloodstream infections is justified, and this new category will have obvious implications for choices about empirical therapy and infection-control surveillance.