Background:  In HIV-positive (HIV⁺ patients (pts), bacteremia (BSI) can have an incidence up to 18%, with mortality up to 30%. Some authors have detected a decline in incidence with the availability of HAART. We describe the epidemiology of BSI among hospitalized HIV⁺  pt.

Methods:  Demographic, clinical, and laboratory data were collected prospectively in consecutive admissions of HIV⁺ pt between September 1999 and December 2003. Case-control studies of risk factors for BSI and association of BSI with mortality were performed; controls were selected based on a random sample of unaffected patients.

Results:  Among 4281 HIV⁺ admissions, 43% were using HAART, CD4 counts were < 200 in 61% and < 50 in 39%, and BSI occurred in 214 (5%) with a mortality of 6%. The most common organisms were (cases, incidences):  S. aureus (61, 1.4%), S. pneumoniae (53, 1.2%), E. coli (22, 0.5%), and P. aeruginosa (10, 0.2%). BSI sources were lung (32%), intravascular catheter (18%,  9% dialysis catheter, 9% other line, urine (6%), skin/soft tissue (5%), and unknown (25%). The patients were 81% African American, 12% Hispanic, 71% male, and 29% female. Median values for age, CD4 count, and viral load were 41, 91, and 37,000, respectively. Cases and controls were well matched on ethnicity, gender, age, and CD4 counts. On univariate analyses (odds ratio [OR], 95% confidence interval [CI]), illicit drug use, OR 1.37, CI 0.99 to 1.90; viral load < 1000, 0.61, 0.39 to 0.97; admission over the prior 3 months 1.33, 0.97 to 1.83; diabetes (DM) 2.00, 1.08 to 3.68; and hypertension (HTN) 1.64, 1.08 to 2.49 were associated factors. Following multivariate logistic regression, significant factors were viral load < 1000 0.58, 0.36 to 0.93]; admission in the prior 3 months 1.40, 1.01 to 1.94]; DM 1.89, 0.99 to 3.56; and HTN 1.53, 0.9 to 2.38. No association was found between BSI and CD4 count, HAART, race, sex, or age. The OR of death was 9.0 (3.41 to 23.5) in patients with BSI controlling for other associated factors (presence of an opportunistic infection, race, sex, and presence of DM or cancer).

Conclusions:  BSI in hospitalized HIV⁺ pt carried a 9-fold increased risk for death. Recent hospital admission was more common in pt with BSI, possibly related to prior invasive procedures (i.e., catheters) or a higher level of debility. Low viral load was protective against BSI and may be a marker for recent, but not current, HAART use. DM and HTN trended toward increased risk of BSI, and emphasize that increases in underlying medical conditions with the aging of the HIV⁺ cohort may lead to increased morbidity.

Keywords: Bacteremia; Mortality; AIDS