Circulating Inflammatory Mediators in Patients with Fever: Predicting Bloodstream Infection

Abstract

ABSTRACT The systemic host response to microbial infection involves clinical signs and symptoms of infection, including fever and elevated white blood cell (WBC) counts. In addition, inflammatory mediators are released, including activated complement product C3a, interleukin 6 (IL-6), and the acute-phase reactant secretory phospholipase A2 (sPLA2). To compare the value of the latter with the former in predicting (the degree of) microbial infection at the bedside, we determined clinical variables and took blood samples daily for 3 consecutive days in 300 patients with a new fever (>38.0°C rectally or >38.3°C axillary). Microbiological culture results for 7 days after inclusion were collected. Patients were divided into clinical and microbial categories: those without and with a clinical focus of infection and those with negative cultures, with positive local cultures or specific stains for fungal (n = 13) or tuberculous infections (n = 1), and with positive blood cultures, including one patient with malaria parasitemia. The area under the curve (AUC) of the receiver operating characteristic (ROC) for prediction of positive cultures was 0.60 (P < 0.005) for peak temperature and 0.59 (P < 0.01) for peak WBC count, 0.60 (P < 0.005) for peak C3a, 0.63 (P < 0.001) for peak IL-6, and 0.61 (P < 0.001) for peak sPLA2. The AUC under the ROC curve for prediction of positive blood cultures was 0.68 (P < 0.001) for peak temperature and 0.56 for peak WBC count (P < 0.05). The AUC for peak C3a was 0.69, that for peak IL-6 was 0.70, and that for sPLA2 was 0.67 (for all, P < 0.001). The degree of microbial invasion is thus a major determinant of the clinical and inflammatory host response in patients with fever. Moreover, circulating inflammatory mediators such as C3a and IL-6 may help to predict positive blood cultures, together with clinical signs and symptoms of the host response to microbial infection, even before culture results are available. This may help in the designing of entry criteria for therapeutic intervention studies.