Overprediction of Carbapenemase-Containing Isolates by an Automated AST Instrument’s Computer Algorithm

Abstract Background The VITEK® 2 is generally accepted as a reliable method for predicting antibiotic resistance mechanisms including aminoglycoside phenotypes, beta lactam phenotypes, impermeability, and penicillinases. However, when it comes to predicting carbapenemases, the research that has been done is inconsistent in both methods and results. Methods We compared the predictions of the VITEK 2 and Advanced Expert System™ (AES) with the results of the modified carbapenemase inactivation method in an academic medical center lab to evaluate the clinical reliability of the VITEK 2 and AES in routine workflow for the detection of carbapenemases. Results Our findings show that the positive predictive value of carbapenemase detection on the VITEK2 and AES is 30.3% (91/300) in our patient population. Conclusions In light of these results, we propose that the VITEK 2 and AES be used as a screening tool for carbapenemase-containing organisms rather than as a definitive test.