Diagnostic concordance between BioFire® FilmArray® Pneumonia Panel and culture in patients with COVID-19 pneumonia admitted to intensive care units: the experience of the third wave in eight hospitals in Colombia

Abstract

Abstract Background The detection of coinfections is important to initiate appropriate antimicrobial therapy. Molecular diagnostic testing identifies pathogens at a greater rate than conventional microbiology. We assessed both bacterial coinfections identified via culture or the BioFire® FilmArray® Pneumonia Panel (FA-PNEU) in patients infected with SARS-CoV-2 in the ICU and the concordance between these techniques. Methods This was a prospective study of patients with SARS-CoV-2 who were hospitalized for no more than 48 h and on mechanical ventilation for no longer than 24 h in 8 ICUs in Medellín, Colombia. We studied mini-bronchoalveolar lavage or endotracheal aspirate samples processed via conventional culture and the FA-PNEU. Coinfection was defined as the identification of a respiratory pathogen using the FA-PNEU or cultures. Serum samples of leukocytes, C-reactive protein, and procalcitonin were taken on the first day of intubation. We analyzed the empirical antibiotics and the changes in antibiotic management according to the results of the FA-PNEUM and cultures. Results Of 110 patients whose samples underwent both methods, FA-PNEU- and culture-positive samples comprised 24.54% versus 17.27%, respectively. Eighteen samples were positive in both techniques, 82 were negative, 1 was culture-positive with a negative FA-PNEU result, and 9 were FA-PNEU-positive with negative culture. The two bacteria most frequently detected by the FA-PNEU were Staphylococcus aureus (37.5%) and Streptococcus agalactiae (20%), and those detected by culture were Staphylococcus aureus (34.78%) and Klebsiella pneumoniae (26.08%). The overall concordance was 90.1%, and when stratified by microorganism, it was between 92.7 and 100%. The positive predictive value (PPV) was between 50 and 100% and were lower for Enterobacter cloacae and Staphylococcus aureus. The negative predictive value (NPV) was high (between 99.1 and 100%); MecA/C/MREJ had a specificity of 94.55% and an NPV of 100%. The inflammatory response tests showed no significant differences between patients whose samples were positive and negative for both techniques. Sixty-one patients (55.45%) received at least one dose of empirical antibiotics. Conclusions The overall concordance was 90.1%, and it was between 92.7% and 100% when stratified by microorganisms. The positive predictive value was between 50 and 100%, with a very high NPV.