To investigate the epidemiology and environmental sources of Fusarium infections in patients with cancer.

Retrospective case-control study conducted following surveillance environmental cultures and DNA analysis of isolated organisms.

A tertiary-care, university cancer center.

In 1996 and 1997, environmental cultures were performed on air samples and water systems. A retrospective chart review was performed for 70 patients with cancer identified with fusariosis between 1987 and 1997. Patients with fusariosis were compared with 49 uninfected control patients who occupied hospital rooms with positive environmental cultures for Fusarium. With the use of random amplification of polymorphic DNA organisms isolated from infected patients were compared with environmental organisms.

Most of the patients with Fusarium (40, 57%) were infected on or within 3 days of admission, indicating community rather than nosocomial acquisition. Patients were comparable in terms of underlying immunocompromised status to 49 uninfected control patients. However, the duration from admission to infection in the patients with fusariosis tended to be shorter than the duration from admission to discharge in the exposed control patients (P = .06). Water cultured from the hospital tanks and from sinks and water fountains was negative for Fusarium. With the use of polymerase chain reaction, environmental isolates did not match clinical ones. Quantitative air sampling showed that the quantitative outdoor Fusarium levels were eightfold higher than the indoor levels. During the rainy summer season, outdoor air concentrations of Fusarium were at their highest, coinciding with the peak incidence of fusariosis at our center.

The most likely source of fusariosis was the external environment rather than nosocomial sources, such as water.