Design and Aim
This systematic review and meta-analysis aims to assess and summarise and evaluate the evidence for the diagnostic value of serum and sputum biomarkers in the differentiation of bacterial versus non-bacterial acute exacerbations of COPD (AECOPD).
Cross-sectional, cohort and randomised controlled studies that describe associations between serum or sputum molecular or cellular biomarkers and evidence of bacterial infection in people with acute exacerbation of COPD were eligible for inclusion. We excluded animal studies, publications of abstracts only, case reports, letters, comments and reviews, and publications in languages other than English. Our review protocol was not published.
Information Sources and Search Strategy
We searched Embase and Medline from inception (1947 for Embase, 1946 for Medline) until 19th March 2020 using a search strategy that included terms for COPD, inflammation & inflammatory markers, bacterial infection and exacerbation (Additional file 1: figure S1). The results were screened for duplications, the inclusion/exclusion criteria applied, and the remaining titles & abstracts were screened for relevancy, requesting full papers where necessary. Uncertainties were screened by a second reviewer and resolved through discussion.
The main outcome was the biomarker concentration in participants defined as having a bacterial exacerbation compared with those defined as having a non-bacterial exacerbation. Where provided, we also extracted the test characteristics (sensitivity, specificity, positive predictive value, negative predictive value) at various biomarker cut points.
Risk of Bias and Assessing the Quality of the Evidence
Risk of bias was assessed using the QUADAS-2 tool. This tool uses four key domains (patient selection, index test, reference standard, flow and timing) to evaluate the risk of bias, with the first three domains also used to evaluate applicability. Signalling questions are used to aid judgements of risk of bias and applicability. The overall quality of the evidence was then assessed qualitatively, with size of cohort, selection criteria, risk of bias, definitions of COPD + exacerbation + bacterial exacerbation and data completeness all contributing to this assessment.
The following data were extracted from the full-text versions of included papers: year published; country in which the study was performed; setting; how the population was defined; how COPD was defined; how exacerbation was defined; number of participants and data completeness; media of samples; biomarkers studied; definition of bacterial exacerbation; results, including average concentrations of biomarkers and suggested cut-offs.
The data were synthesised descriptively, with each biomarker being assessed individually. For biomarkers where there were sufficient data, mean biomarker concentration in patients with and without evidence of bacterial infections were compared using random-effects meta-analysis estimated using restricted maximum likelihood, with absolute mean differences calculated and the results displayed using forest plots. Stata (v16.1) software was used for the meta-analysis.
BMC Pulm Med. 2022;22(194) © 2022 BioMed Central, Ltd.