Total parenteral nutrition-related infections. Prospective epidemiologic study using semiquantitative methods

Studies of total parenteral nutrition-related infection have incorrectly relied on positive results on culture of the cannula tip to confirm the source. We undertook a prospective study of total parenteral nutrition-related infections in adult patients by obtaining blood from all total parenteral nutrition lines for pour-plate culture twice weekly and culturing intravascular line segments by the technique of Maki. Twelve of 100 courses of total parenteral nutrition (12 percent) in 69 patients resulted in infections--five (5.0 percent) had sepsis, and seven (7.0 percent) had local infection. In five of these 12, pour-plate culture gave positive results (five of 38 pour plates) with counts of 8 colony-forming units per ml (cfu/ml) for Candida tropicalis, and 25 to more than 1,000 for bacterial isolates. In nine of 12, culture of the intravascular line segment gave positive results with more than 50 cfu/ml. Pathogens isolated from intravascular line segments included Staphylococcus epidermidis (three cases), Candida species (three cases), Staphylococcus aureus (two cases), Serratia marcescens (one case) and mixed bacterial pathogens (one case). In contrast, pour-plate culture gave positive results in only seven of 88 uninfected (control) courses (318 pour plates), and culture of intravascular line segments gave positive results in two of 65 uninfected courses (p less than 0.001). No differences existed among patients with and without infection with regard to age, underlying disease, surgery, systemic antibiotic usage, or the presence of other infections. The duration of total parenteral nutrition was longer in courses without infection than in courses with infection (14.7 +/- 9.4 days versus 11.0 +/- 4.0 days; p less than 0.02). In six of 12 courses with infection, the line had been violated compared with 22 of 88 courses without infection (p less than 0.001). T-connectors for the centra administration of intralipid were associatd with infection (p less than 0.02). The value of routine pour-plate culture was illustrated in three courses in which the positive pour-plate culture results antedated positive blood culture results or line removal.