Pre-emptive fluconazole (fcz) anti-fungal therapy is often based upon Candida colonisation of at least 2 non-contiguous non-sterile sites. The aim of this study was to evaluate the relationship between candidaemia and prior colonisation of non-sterile sites.

A retrospective observational study was performed in the intensive care unit/high dependency unit (ICU/HDU) of a University hospital on alternate years from 1999–2007, where a pre-emptive anti-fungal therapy policy was introduced in 2005.

A higher proportion of blood isolates were Candida glabrata compared with non-sterile isolates (16/46 vs 106/1062; p < 0.001), similarly a greater proportion of blood isolates were fcz-resistant compared with non-sterile isolates (15/46 vs 101/1062; p < 0.001). No trend over time was detected in the proportion of C. glabrata and Candida albicans isolates from blood and non-sterile sites, or in the fcz-sensitivity of isolates from these sites. C. glabrata candidaemia was more likely to occur in the absence of non-sterile site colonisation compared with non-glabrata candidaemia (12/16 vs 8/30; p = 0.005). Of candidaemic patients, 43% had no preceding colonisation by any Candida spp.; in 67% of these patients, candidaemia was due to C. glabrata.

Pre-emptive therapy based upon colonisation of at least two sites may be inadequate as 43% of candidaemic patients had no evidence of prior colonisation, 67% of whom had candidaemia due to C. glabrata. Furthermore if pre-emptive anti-fungal therapy is instituted in non-colonised patients there is a risk of selecting an inappropriate anti-fungal for C. glabrata. Despite the introduction of pre-emptive fcz therapy, no time trend was detected in the proportion of fcz-sensitive isolates from blood and non-sterile sites.