There continues to be difficulty making the clinical distinction betwe
en fungal colonization and systemic infection in critically ill surgic
al patients. This distinction is important, given the potential risks
of aggressive antifungal therapy. In order to evaluate the significanc
e of fungal infections by various sites, we retrospectively reviewed t
he clinical courses of patients with cultures positive for fungi (Cand
ida species or Torulopsis glabrata) in the SICU of LAC + USC Medical C
enter from January 1992-December 1993. There were 129 patients who wer
e culture positive for Candida (110 patients) or Torulopsis glabrata (
19 patients). There were 187 positive cultures. Fifty-five patients (4
3%) had systemic fungal infections (two or more sites, or fungemia). T
he proportion of patients with positive cultures from any given site g
oing on to develop systemic infections was similar (wound, 49 per cent
; urine, 54 per cent; sputum, 57 per cent; drain, 68 per cent; P = 0.6
1, NS). The mortality for SICU patients with systemic fungal infection
was significantly increased (36.3% versus 10.5%, P < 0.05) when compa
red with SICU patients in general. No significant increase in mortalit
y was seen in patients with single site isolation (13.5% versus 10.5%,
P = 0.52). This study suggests that although systemic fungal infectio
n is associated with increased mortality in SICU patients, no single s
ite of isolation is superior to others in predicting which patients ar
e likely to develop systemic infection. Prospective studies with antif
ungal agents with reduced toxicity are justified in patients with sing
le site isolation.