Longitudinal changes in procalcitonin in a heterogeneous group of critically ill patients

Abstract

Objective: The utility of procalcitonin for the diagnosis of infection in the critical care setting has been extensively investigated with conflicting results. Herein, we report procalcitonin values relative to baseline patient characteristics, presence of shock, intensive care unit time course, infectious status, and Gram stain of infecting organism.

Design: Prospective, multicenter, observational study of critically ill patients admitted to intensive care unit for >24 hrs.

Setting: Three tertiary care intensive care units.

Patients: All consenting patients admitted to three mixed medical-surgical intensive care units. Patients who had elective surgery, overdoses, and who were expected to stay <24 hrs were excluded.

Interventions: Patients were followed prospectively to ascertain the presence of prevalent (present at admission) or incident (developed during admission) infections and clinical outcomes. Procalcitonin levels were measured daily for 10 days and were analyzed as a function of the underlying patient characteristics, presence of shock, time of infection, and pathogen isolated.

Main results: Five hundred ninety-eight patients were enrolled. Medical and surgical infected cohorts had similar baseline procalcitonin values (3.0 [0.7-15.3] vs. 3.7 [0.6-9.8], p=.68) and peak procalcitonin (4.5 [1.0-22.9] vs. 5.0 [0.9-16.0], p=.91). Infected patients were sicker than their noninfected counterparts (Acute Physiology and Chronic Health Evaluation II 22.9 vs. 19.3, p<.001); those with infection at admission had a trend toward higher peak procalcitonin values than did those whose infection developed in the intensive care unit (4.9 vs. 1.4, p=.06). The presence of shock was significantly associated with elevations in procalcitonin in cohorts who were and were not infected (both groups p<.003 on days 1-5).

Conclusions: Procalcitonin dynamics were similar between surgical and medical cohorts. Shock had an association with higher procalcitonin values independent of the presence of infection. Trends in differences in procalcitonin values were seen in patients who had incident vs. prevalent infections.