Clinical sign and biomarker-based algorithm to identify bacterial pneumonia among outpatients with lower respiratory tract infection in Tanzania

Abstract

Background: Inappropriate antibiotics use in lower respiratory tract infections (LRTI) is a major contributor to resistance. We aimed to design an algorithm based on clinical signs and host biomarkers to identify bacterial community-acquired pneumonia (CAP) among patients with LRTI.

Methods: Participants with LRTI were selected in a prospective cohort of febrile (≥ 38 °C) adults presenting to outpatient clinics in Dar es Salaam. Participants underwent chest X-ray, multiplex PCR for respiratory pathogens, and measurements of 13 biomarkers. We evaluated the predictive accuracy of clinical signs and biomarkers using logistic regression and classification and regression tree analysis.

Results: Of 110 patients with LRTI, 17 had bacterial CAP. Procalcitonin (PCT), interleukin-6 (IL-6) and soluble triggering receptor expressed by myeloid cells-1 (sTREM-1) showed an excellent predictive accuracy to identify bacterial CAP (AUROC 0.88, 95%CI 0.78-0.98; 0.84, 0.72-0.99; 0.83, 0.74-0.92, respectively). Combining respiratory rate with PCT or IL-6 significantly improved the model compared to respiratory rate alone (p = 0.006, p = 0.033, respectively). An algorithm with respiratory rate (≥ 32/min) and PCT (≥ 0.25 μg/L) had 94% sensitivity and 82% specificity.

Conclusions: PCT, IL-6 and sTREM-1 had an excellent predictive accuracy in differentiating bacterial CAP from other LRTIs. An algorithm combining respiratory rate and PCT displayed even better performance in this sub-Sahara African setting.

Keywords: Bacterial community-acquired pneumonia; Biomarkers; PCT; Predicting algorithm.

Fig. 1

Study flow chart. LRTI lower respiratory tract infection

Fig. 2

Aetiologies distribution of community-acquired pneumonia (n = 32). Of note, no Mycoplasma pneumoniae, Chlamydia pneumoniae or Legionella pneumophila were identified

Fig. 3

Plasma concentration of immune and endothelial dysfunction markers at clinical presentation according to the diagnosis. Boxplot with median and interquartile range. Concentrations reported in pg/mL except CRP in mg/L. P values were computed using the Wilcoxon-Mann Whitney test and were adjusted for multiple comparisons using Bonferroni method. P * < 0.05; ** < 0.01; *** < 0.001. CAP, community-acquired pneumonia

Fig. 4

Accuracy of markers of endothelial and immune activation, measured in adults presenting with clinical lower respiratory tract infection to outpatient clinics in predicting bacterial community-acquired pneumonia. Nonparametric ROC curves were generated and AUROC were plotted to illustrate the ability of these markers to discriminate between bacterial community-acquired pneumonia and other lower respiratory tract infection. AUROCs for the outcome of each marker are presented to the right of its respective forest plot, with 95% CIs in parentheses. Angpt-1 angiopoietin-1, Angpt-2 angiopoietin-2, AUROC area under the receiver operating characteristic, CHI3L1 chitinase-3-like protein-1, CI confidence interval, CRP C-reactive protein, IL-6 interleukin-6, IL-8 interleukin-8, IP-10 interferon-gamma-inducible protein-10, PCT procalcitonin, ROC receiver operating characteristic, sICAM-1 soluble intercellular adhesion molecule-1, sTNFR-1 soluble tumour necrosis factor receptor-1, sTREM-1 soluble triggering receptor expressed on myeloid cells, sVCAM-1 soluble vascular cell adhesion molecule-1, sVEGFR1 soluble variant of vascular endothelial growth factor receptor 1

Fig. 5

Classification and regression tree analysis to predict bacterial community-acquired pneumonia in patients presenting with lower respiratory tract infection at outpatient clinics in Tanzania. a All variables (vital signs and biomarkers) were added to the model. b Forced first respiratory rate and PCT were added to the model. c Forced first respiratory rate and PCT cut-off 0.25 µg/l were added to the model. For all models, the cost of misclassifying a patient that had bacterial community-acquired pneumonia as 10 times the cost of misclassifying patients that had other lower respiratory tract infection. Cut points selected by the analysis are indicated between the parent and child nodes. Below each terminal node, the predicted categorization for those patients is indicated. Algorithm performance characteristics are presented in Table 3. PCT procalcitonin