The CPDX rapid screening accuracy was confirmed by the proportion of categorical agreement for CPDX rapid screening accuracy using VME, ME, and mE which was similar to the RAST accuracy [13]. Therefore, this implies that CPDX rapid screening is a useful method for the early detection of third-generation cephalosporin-resistant bacteria in E. coli and K. pneumoniae bacteremia. When detecting resistance genes that result in third-generation cephalosporin resistance in E. coli and K. pneumoniae, AmpC cannot be detected using Verigene or other commonly used genotypic testing. In this study, third-generation cephalosporin resistance due to AmpC production was detected, in only one case using CPDX rapid screening. Therefore, in E. coli and K. pneumoniae bacteremia, CPDX rapid screening may be more sensitive for third-generation cephalosporin resistance. At the time of the introduction of this method, standardized methods for antimicrobial susceptibility testing directly from blood culture, such as RAST published by EUCAST, had not been published. This method was developed for easy and rapid screening of third-generation cephalosporin-resistant bacteria at our hospital and is not standardized. However, CLSI M100 ED32 published a test method for direct disk diffusion from positive blood cultures to determine susceptibility to ceftriaxone and ceftazidime 8–10 h after incubation in Enterobacterales [9]. CPDX rapid screening differs from the CLSI M100 procedure for testing disk diffusion directly from positive blood culture broth in that the incubation time is 6 h, one drop of blood culture broth is dispensed on Chocolate Agar EX II with vancomycin, streaked with an inoculating loop, and CPDX disks are used. Although this method was shown to be similar in accuracy to RAST, it is necessary to follow the CLSI M100 procedure for the rapid detection of resistant organisms using a standardized method. Although RAST requires MALDI-TOF MAS for the identification of bacterial species, the CLSI M100 procedure does not require identification of bacterial species. Therefore, by following the CLSI M100 procedure, the test results can be reported as a standardized method without the need to introduce equipment such as MALDI-TOF MS, which is expected to be introduced at many facilities.

The primary outcome of this study was that the proportion of cases wherein the administration of susceptible antimicrobial agents was modified within 24 h of blood culture -positive reports, significantly increased, despite 67.9% of cases reporting the results of CPDX rapid screening. Rapid phenotypic susceptibility testing has reduced time‐to‐appropriate antibiotic therapy [14, 15, 22], and this study showed similar results. Therefore, intervention with AST using CPDX rapid screening may contribute to appropriate antimicrobial use.

In E. coli and K. pneumoniae, most third-generation cephalosporin-resistant strains are ESBL-producing bacteria [1], and the first-line agents for ESBL-producing bacteria are carbapenems [23]. In the CPDX screening group, carbapenems were selected in most cases that were modified to a susceptible antimicrobial agent, suggesting that an appropriate antimicrobial agent was selected. However, several cases have been reported for cephamycins. Cephamycins are not recommended as a treatment for ESBL-producing bacterial infections [23]; however, several retrospective cohort studies comparing them to carbapenems have demonstrated their effectiveness [24,25,26]. Therefore, cephamycins are often used in Japan for urinary tract infections caused by ESBL-producing bacteria, and cephamycin has been selected for several cases. Thus, cephamycins and oxacephems, which are considered effective against bacteremia caused by ESBL-producing bacteria, are available in Japan and are administered as an alterative to carbapenems. In this study, 96.5% of the cases in which CPDX rapid screening was performed were cefmetazole susceptible. The use of an antimicrobial prescription protocol when third generation cephalosporin resistance is determined by the CPDX screening results used in this study may increase the use of carbapenems. However, in non-severe cases of uncomplicated urinary tract infections, in which the causative organisms are mostly E. coli, a protocol that suggests prescribing cephamycins or oxacephems may reduce the requirement for carbapenems prescriptions.

In the CPDX screening group, the method of identification of bacterial species was changed to MALDI-TOF MS during the study period. The introduction of MALDI-TOF MS increases the proportion of appropriate empirical treatments [27,28,29,30] and may affect the primary outcome. In these studies, the analysis included bacterial species other than E. coli and K. pneumoniae, and in E. coli and K. pneumoniae bacteremia, there were no cases [28] or a few cases [29, 30] of modification to carbapenems, the first-line agents in infections caused by ESBL-producing bacteria [23]. Therefore, MALDI-TOF-MAS is considered to have a small impact on the appropriate empirical treatment for ESBL-producing bacteremia, and its impact on the primary outcome in this study is considered to be small.

This study did not show improvement in secondary outcomes, such as 30-day mortality, hospital mortality, length of hospital stays, and cost of intravenous antibiotic therapy. Tumbarello M et al. reported that inappropriate empiric therapy increased mortality compared to appropriate empiric therapy in bacteremia caused by ESBL-producing bacteria (59.5% vs. 18.5%) [8]. In RCTs examining the clinical impact of rapid genotypic susceptibility testing and rapid phenotypic susceptibility testing, the highest mortality was 12.3%, and these reports did not show an improvement in mortality [14,15,16,17]. In terms of mortality improvement, this effect is expected only in the patient population with high mortality. Therefore, no improvement in mortality was observed in this study. In addition, these studies have not shown an improvement in the length of hospital stays [14,15,16,17]. The effect of the methods for determining antimicrobial resistance on the day of positive blood culture on improving the length of hospital stays is considered to be small. Moreover, the cost of intravenous antibiotic therapy could not be improved because the length of hospital stays did not improve.

This study had several limitations. First, this was a retrospective study, and although we had confirmed that there were no significant differences in the characteristics of the patients, we may not know all the factors affecting the outcomes. Second, the setting of each group was divided by time period; factors other than CPDX rapid screening may have affected the outcomes. The introduction of MALDI-TOF MS increases the proportion of appropriate empirical treatments [28,29,30], and the change to MALDI Biotyper for bacterial identification during the study period may have had any impact on the primary outcome.

This is the primary study providing insights into the utility and accuracy of CPDX rapid screening. Although this study showed that using CPDX rapid screening for the early detection of third-generation cephalosporin-resistant E. coli and K. pneumoniae could contribute to appropriate antimicrobial use, prospective studies using screening tests based on standard methods published by the CLSI are needed.