To determine whether the conventional rate for central venous catheter (CVC)-associated bloodstream infection (BSI) accurately reflects risk for patients exposed for a variety of in situ periods.

Intensive care unit patients (n = 1,375) were monitored for 7,467 CVC-days. They were monitored until catheter removal, until diagnosis of CVC-associated BSI, or for 24 hours after discharge.

The BSI rate was 3.7 per 1,000 CVC-days. Ninety-three percent of these patients had CVCs in situ for 1-15 days. These patients were exposed to 59.7% of all CVC-days; the remaining 7% were exposed to 40.3% of all CVC-days. BSI rates stratified by exposure periods of 1-5 and 6-15 days were 2.1 and 4.5 per 1,000 CVC-days, respectively. The rates for 16-30 and 31-320 days were 10.2 and 2.1 per 1,000 CVC-days, respectively. The probability of BSI with a CVC in situ was 6 in 100 by day 15, 14 in 100 by day 25, 21 in 100 by day 30, and 53 in 100 by day 320.

The conventional aggregated rate better reflects the risk for the majority of patients rather than for patients exposed to the majority of CVC-days. It does not reflect the true probability of risk for all exposures, especially beyond 30 days. CVCs in situ from 1 to 15 days had less risk of BSI than CVCs in situ more than 15 days, which may explain why scheduled CVC replacement at days 5 to 7 has not been found beneficial.