Performing Reflexive Toxin A/B Enzyme Immunoassay in a Two-Step Algorithm for the Diagnosis of Clostridium difficile Infection Has Limited Clinical Utility

The differentiation of Clostridium difficile infection (CDI) from colonization is challenged by the suboptimal clinical specificity of nucleic acid amplification tests (NAAT). In this study, we examined the utility of testing for toxin via enzyme immunoassay (EIA) in specimens already tested by NAAT for the diagnosis of CDI in an attempt to differentiate colonization from infection. We tested 59 stool samples for the presence of C. difficile toxin B gene by NAAT followed by EIAs for glutamate dehydrogenase (GDH EIA) and toxins A and B (Toxin EIA). Two infectious disease physicians independently reviewed the patients electronic medical records retrospectively to categorize each patient as CDI-Likely, CDI-Unlikely, or CDI-Indeterminate. Clinical sensitivities and specificities were calculated using 3 definitions of \"true\" CDI status, being: (1) concordance between both reviewers, (2) concordance, and CDI-Indeterminate/discordant cases classified as CDI-Likely, and (3) concordance, and CDI-Indeterminate/discordant cases classified as CDI-Unlikely. Based on these definitions, clinical sensitivity and specificity for NAAT was 100% and 49-94%, GDH EIA was 83-85% and 43-89%, and Toxin EIA was 39-42% and 83-100%, respectively. 85% (22 of 26) of patients who were NAAT-positive but Toxin EIA-negative symptomatically benefited from treatment for CDI. The addition of EIA to NAAT for CDI diagnosis had limited utility for differentiating colonization from CDI and could have led to under treatment of patients with CDI.