To assess the added value of surveying patients after discharge from the intensive care unit.
Prospective cohort study.
Medical intensive care unit of a large teaching hospital.
All patients admitted to the intensive care unit for 48 hrs or more from October 1995 to November 1997.
We prospectively surveyed 1,068 patients during their intensive care unit stay and for 5 days after intensive care unit discharge. We detected 554 intensive care unit-acquired infections, yielding an infection rate of 70.7 per 1,000 patient days. Of these, only 31 infections (5.6%) in 27 patients were detected after intensive care unit discharge. If postdischarge surveillance was targeted on patients who had had a central vascular catheter while in the intensive care unit, only one infected patient would have been missed, but only 554 out of 889 would have been followed up (sensitivity, 96.2%; specificity, 38.7%; negative predictive value, 99.7%).
Surveillance of all patients discharged from the medical intensive care unit is not recommended, as it is resource demanding and allows the detection of few additional infections. However, targeted postdischarge surveillance could be a rational alternative, and selection criteria need to be refined and validated.