This study, conducted at a single tertiary care hospital in the Republic of Korea, showed that the direct stool Xpert Carba-R assay had favorable diagnostic performance for detecting rectal CPE colonization. A moderate level of concordance was observed between results from direct stool assay and those from corresponding clinical isolates, in a setting where the Klebsiella pneumoniae and blaKPC were the most frequently identified species and gene, consistent with national surveillance data from the Korean Disease Control and Prevention Agency (KDCA) between 2019 and 2023 [15]. Notably, blaKPC showed the highest PPV of 90.8%, while blaIMP-1 showed a PPV of 0%, as none of the blaIMP-1 detection from direct stool assay were confirmed in corresponding isolates.

Our infection control approach for CPE/CRE surveillance was influenced by healthcare reimbursement policies. The Korean National Health Insurance only reimburses single-room isolation for culture-confirmed CPE/CRE cases. For patients who test positive by direct stool Xpert assay but negative by culture, single-room isolation costs approximately ten times more than contact precautions alone, creating a significant financial burden for patients. Therefore, we implemented contact precautions without single-room isolation for these cases as an interim measure until culture results became available. While this represents a compromise in infection control to immediate single-room isolation based on molecular results, it reflects real-world healthcare system constraints that many institutions face when implementing rapid molecular diagnostics for CPE screening.

Globally, the reported incidence of CRE has increased markedly, and the disease has been designated as part of the critical group in the World Health Organization Bacterial Priority Pathogen List [16]. In this context, rapid diagnostic assays can serve a pivotal role in the active surveillance of CPE, particularly by enabling early detection through their reduced turnaround time. Park et al. reported a single-center retrospective study evaluating the effect of enhanced CPE screening using the Xpert Carba-R assay. Following its implementation, significant reductions were observed in the median duration of CPE exposure (10.8 days vs. 1 day), the number of contacts per CPE case (11 vs. 1), and the incidence of hospital-acquired CPE (1.9 vs. 1.1 per 1,000 admissions). These improvements were attributed to the shortened turnaround time and active surveillance enabled by applying the Xpert Carba-R assay directly to rectal swabs [17]. Similarly, our study demonstrated that the direct stool Xpert Carba-R assay provided a shorter turnaround time than culture-based Xpert Carba-R assay, along with favorable diagnostic performance.

Our study demonstrated relatively high PPV for blaKPC; however, blaIMP-1 showed a PPV of 0%, indicating complete lack of concordance. Additionally, detection of multiple carbapenemase genes in direct stool assay frequently did not correspond with the presence of multiple genes in the clinical CPE isolates, with none of the multigene detection from direct stool assay being confirmed in cultured isolates. Within our hospital`s surveillance workflow, the direct stool Xpert Carba-R assay yielded three apparent false-negative results. These findings highlight potential limitations of relying solely on the direct stool Xpert Carba-R assay for CPE detection, despite its advantages.

As shown in our study, several factors should be considered when interpreting the results of the direct stool Xpert Carba-R assay. First, high PPV observed for blaKPC might be attributed to the high prevalence of blaKPC in the Republic of Korea. According to national surveillance data from 2023, KPC accounted for 77.4% of CPE isolates, followed by NDM (15.9%) and OXA-48-like enzymes (6.2%) [15]. Because PPV is influenced by prevalence, the regional prevalence of specific carbapenemase genes should be considered when planning proactive surveillance using the direct stool Xpert Carba-R assay.

Second, the detection of multiple carbapenemase genes from direct stool assay has been frequently reported in previous studies. For example, a multicenter study from the Gulf Cooperation Council found that 128 of 529 (26.1%) rectal swab specimens harbored multiple carbapenem resistance genes [18]. However, this study did not evaluate the corresponding carbapenemase genes from clinical isolates. Another study reported two cases of multiple gene detection (blaKPC/blaIMP-1 and blaKPC/blaOXA-48) during proactive ICU surveillance; however, only blaKPC was confirmed in the two cases from CRE-targeted culture [12]. In contrast, another study reported that more than 90% of multiple carbapenemase gene detections were confirmed in clinical isolates, where culture was conducted without restricting to CRE-targeted methods [19]. In our study, 11 cases showed multiple gene detection in direct stool assay, suggesting that detection of multiple carbapenemase genes from direct stool samples rarely represents colonization with single CPE co-producing multiple enzymes. Rather, it might reflect concurrent colonization with other carbapenemase-producing pathogens, such as Pseudomonas or Acinetobacter species. Furthermore, another study has demonstrated the simultaneous isolation of multiple strains of the same Enterobacterales species, each harboring a different carbapenemase gene [20], providing an alternative explanation for these findings. Although multiple gene detection by the direct stool Xpert Carba-R assay should be interpreted with caution, our findings suggest a high probability of CPE colonization in such cases.

Third, in our study, none of the blaIMP-1 genes detected by the direct stool Xpert Carba-R assay were confirmed in the corresponding clinical isolates, with eight showing no carbapenemase genes and one showing blaKPC instead. This finding is consistent with previous reports that have described false-positive blaIMP-1 gene results in the context of CPE colonization [12, 13, 21]. Such discrepancies can be explained by colonization with other carbapenemase-producing Gram-negative bacilli (CRGNB). Although the Xpert Carba-R assay is designed to detect blaIMP-1, it can also identify various IMP subgroups without distinction [12]. In the Republic of Korea, IMP-6 is the most prevalent carbapenemase among Pseudomonas aeruginosa isolates [22, 23]. A previous study also reported that P. aeruginosa was the most frequently isolated organism in patients with false-positive stool Xpert Carba-R blaIMP-1 results [19]. Furthermore, cultures using ChromID CARBA agar may fail to detect CPE with low-level carbapenem resistance [24]. It is possible that multiple CPE strains were initially present in stool samples, but some strains with low-level carbapenem resistance have failed to grow on the selective media. Therefore, although detection of blaIMP-1 by the direct stool Xpert Carba-R assay might raise concerns for CPE colonization, the complete absence of confirmation in culture-based testing suggests that blaIMP-1 detection should be interpreted with extreme caution. Physicians should remain vigilant for possible colonization with other CRGNBs in such cases.

This study has several limitations. First, it was conducted at a single tertiary care center, which can limit the generalizability of the findings to other healthcare settings or regions with differing levels of CPE endemicity and carbapenemase gene distribution. Second, the reference test was conducted using Xpert Carba-R assays on clinical CPE isolates rather than an in-house PCR method. Consequently, carbapenemase genes not targeted by the Xpert Carba-R assay might have been missed, potentially contributing to discordant results. Third, we used MHT for a phenotypic assay. However, as the Xpert Carba-R assay was performed in parallel to overcome the known low sensitivity of the MHT, the impact of this limitation would be minimal. Fourth, because organisms other than CRE were not identified from rectal swab cultures, the applicability of our findings to other CRGNB is limited. Finally, we did not examine Ct values from the Xpert Carba-R assay. Ct value analysis would indeed provide valuable insights into the relationship between bacterial load and culture discordance. This represents a missed opportunity for deeper analysis that could have strengthened our interpretations of discordant results.

In conclusion, the Xpert Carba-R assay serves as a clinically useful molecular assay for rapid detection of CPE from direct stool samples. However, its diagnostic performance varies significantly depending on the specific carbapenemase gene, with particularly poor performance for blaIMP-1 detection, and substantial discrepancies can occur between results from direct stool samples and those from cultured isolates. Therefore, when using the direct stool Xpert Carba-R assay as a proactive screening tool, it is important to recognize these variations in diagnostic performance and potential discordance to inform clinical decision-making and optimize infection control strategies.