Abstract
Urine cultures are among the highest-volume tests in clinical microbiology laboratories and usually require considerable manual labor to perform. We evaluated the APAS Independence automated plate reader system and compared it to our manual standard of care (SOC) for processing urine cultures. The APAS device provides automated image interpretation of urine culture plate growth and sorts those images that require further evaluation. We examined 1,519 specimens over a 4-month period and compared the APAS growth interpretations to our SOC. We found that 72 of the 1,519 total specimens (4.74%) had growth discrepancies, where these specimens were interpreted differently by the APAS and the technologist, which required additional evaluation of plate images on the APAS system. Overall, there were 56 discrepancies in pathogen identification, which were present in 3.69% of the cultures. An additional pathogen was uncovered in a majority of these discrepancies; 12 (21.4%) identified an additional pathogen for the SOC, and 40 (71.4%) identified an additional pathogen for the APAS workflow. We found 214 (2.69%) antimicrobial susceptibility test (AST) discrepancies; 136 (1.71%) minor errors (mEs), 41 (0.52%) major errors (MEs), and 36 (0.45%) very major errors (VMEs). Many of the MEs and VMEs occurred in only a small subset of 13 organisms, suggesting that the specimen may have had different strains of the same pathogens with differing AST results. Given the significant labor required to perform urine cultures, the APAS Independence system has the potential to reduce manual labor while maintaining the identity and AST results of urinary pathogens. IMPORTANCE Urine cultures are among the highest-volume tests performed in clinical microbiology facilities and require considerable manual labor to perform. We compared the results of our manual SOC workflow with that of the APAS Independence system, which provides automated image interpretation and sorting of urine culture plates based on growth. We examined 1,519 urine cultures processed using both workflows and found that only 4.74% had growth pattern discrepancies and 3.69% pathogen identification discrepancies. There was substantial agreement in AST results between workflows, with only 2.69% having discrepancies. Only 1.71% of the ASTs had mEs, 0.52% had MEs, and 0.45% had VMEs, with most of the MEs and VMEs belonging to a small subset of organisms. The APAS system significantly decreased manual urine culture processing, while providing similar results to the SOC. As such, incorporating such automation into laboratory workflows has the potential to significantly improve efficiency.