[Diagnostic and predictive value of critical illness scores and biomarkers in bloodstream infection-associated pneumonia].

Objective: To compare the early and late predictive value of several critical illness scores (CISs) and biomarkers in patients with bloodstream infection (BSI)-associated pneumonia, and to identify the value of procalcitonin (PCT) in etiological diagnosis.

Methods: Patients with at least one positive blood culture within 24 hours admission to department of emergency of China-Japan Friendship Hospital from January 2014 to December 2018 and with final diagnosis of pneumonia were enrolled. Sequential organ failure assessment (SOFA), mortality in emergency department sepsis (MEDS), Logistic organ dysfunction system (LODS), and acute physiology and chronic health evaluation II (APACHE II) scores were calculated based on the first parameters on the day of admission. Differences of various indicators among different Gram-stained bacterial infections and among patients with different prognosis at 28-day or 60-day were compared. Receiver operating characteristic (ROC) curve was used to analyze the value of biomarkers in differential diagnosis of pneumonia caused by single bacterial infection, and the predictive value of several CISs and biomarkers on 28-day and 60-day death of patients with pneumonia.

Results: Among 540 patients with pneumonia caused by single bacterial infection, 256 (47.4%) patients with Gram-positive bacteria (GPB) infection and 284 (52.6%) with Gram-negative bacteria (GNB) infection. The 28-day mortality was 29.4% (159/540) and the 60-day mortality was 36.3% (196/540). PCT level was significantly higher in patients with GNB infection than that in GPB infected patients [μg/L: 1.99 (0.32, 13.19) vs. 0.45 (0.13, 3.53), P < 0.01]. There were significant differences of CISs and biomarkers between death group and survival group in predicting 28-day and 60-day mortality in BSI-associated pneumonia. ROC curve analysis showed that: (1) the optimal cut-off value of PCT in the diagnosis of single bacterial infection was 0.48 μg/L, with the area under ROC curve (AUC) was 0.739 [95% confidence interval (95%CI) was 0.686-0.793]. When PCT value was greater than 4.49 μg/L, the specificity of diagnostic of GNB infection could reach 81.8%, and the positive predictive value (PPV) was 75.0%. When PCT value was greater than 10.16 μg/L, the diagnostic specificity could reach 91.2%. (2) In the prediction of 28-day and 60-day mortality, the SOFA score showed highest AUC [28-day: 0.818 (95%CI was 0.768-0.867), 60-day: 0.800 (95%CI was 0.751-0.849)]. SOFA score greater than 8.5 points could help to predict 28-day and 60-day mortality for pneumonia patients with specificity of 90.5% and 91.6%, respectively. AUC of PCT for predicting 28-day and 60-day mortality in patients with BSI associated with pneumonia was 0.637 (95%CI was 0.575-0.700) and 0.628 (95%CI was 0.569-0.688), respectively. When PCT value was greater than 8.15 μg/L, the specificity and negative predictive value (NPV) were 80.2% and 75.1% respectively, and they could reach 80.2% and 68.7% when PCT value was greater than 7.46 μg/L.

Conclusions: PCT is more reliable in the identification of pathogen type in BSI-associated pneumonia, while CISs may be more advantageous in the assessment of early and late prognosis.