Objective. —To determine risk factors for bloodstream infections (BSIs) in an outbreak among patients receiving home intravenous infusion therapy. Desing. —Case-control and retrospective cohort studies. Setting. —Home health agency. Patients. —Patients receiving home intravenous infusion therapy from Rhode Island Home Therapeutics (RIHT) from January through December 1993. Main Outcome Measure. —Development of primary BSI. Methods. —We compared patients with BSI (ie, case patients) with randomly selected noninfected RIHT patients receiving intravenous therapy, conducted a cohort study of all RIHT patients receiving intravenous therapy via a central venous catheter (CVC), and conducted a culture survey of injection cap luminal fluid. Results. —Case patients were more likely than controls to have had therapy via a CVC (11/11 vs 14/32; odds ratio [OR] undefined; P<.001) or total parenteral nutrition and intralipid therapy (TPN/IL) (9/11 vs 3/32; OR, 43.5; 95% confidence interval [CI], 4.9 to 510.0). Among RIHT patients with CVCs, risk factors for BSI were receipt of TPN/IL (9/35 vs 2/67; rate ratio [RR], 8.6; 95% CI, 2.0 to 37.7) or use of a needleless infusion system (10/41 vs 1/61; RR, 14.9; 95% CI, 2.0 to 111.8). Only the combination of both exposures was significantly associated with development of a BSI (P<.001). Luminal fluid from injection caps of needleless devices was significantly more likely to be culture positive than fluid from protected-needle devices (5/23 vs 0/18; RR undefined; P=.04). Conclusions. —Our data suggest that a needleless device used for TPN/IL was associated with increased risk of BSI when injection caps were changed every 7 days. (JAMA. 1995;273:1862-1864)