Rapid response team-triggered procalcitonin measurement predicts infectious intensive care unit transfers

Objective: Determine if procalcitonin at the time of initial rapid response team activation identifies patients who are likely to need subsequent intensive care unit transfer. Design: Prospective observational cohort study. Setting: Urban, tertiary care hospital with rapid response team activation through an electronic modified early warning score. Patients: One hundred nineteen oncology and 100 consecutive non-oncology patients after initial rapid response team visit precipitated by an elevated electronic modified early warning score were recruited. Rapid response team activations by request of nursing or for other reasons were not studied. Five oncology patients seen by a rapid response team for complications of interleukin-2 therapeutic infusions were subsequently excluded. Interventions: Residual serum from the next ordered clinical test (within 12 hrs) was retrieved, frozen, and stored for procalcitonin determination. A second sample 12-24 hrs after the initial specimen was also retrieved if available and if the patient had not yet been transferred to the intensive care unit. Measurements and Main results: Seventy-three patients (33%) were transferred to the intensive care unit. Rapid response team activations that did not result in intensive care unit transfer had significantly lower procalcitonin levels (median 0.28 ng/mL [interquartile range 0.09-1.24]) than those that resulted in intensive care unit transfer (median 0.51 ng/mL [interquartile range 0.11-1.97], p = .0001) but the area under the receiver operating curve was only 0.656. The change in procalcitonin level in patients with intensive care unit transfers was very heterogeneous but was significantly increased compared to the change in patients not transferred to the intensive care unit. Procalcitonin levels for intensive care unit transfers for probable or definite infection were 2.28 ng/mL [interquartile range 0.68-8.05], and were significantly greater than rapid response team visits that did not result in transfer (p = .0001). The difference between infectious and noninfectious intensive care unit transfers (0.95 ng/mL [interquartile range 0.26-1.89]) was also significant (p = .03). The procalcitonin levels of patients with noninfectious intensive care unit transfers were also different than the levels of patients who never transferred (p = .04). Conclusions: Preliminary results suggest procalcitonin levels in patients at the time of initial visit by a rapid response team correlate with the need for subsequent intensive care unit transfer, particularly for infectious reasons. (Crit Care Med 2012;40:2090-2095)