Table 3 summarized the prevalence of ESBL-producing E. coli according to specific allocation. In ICU, Hsueh et al. conducted an analysis of clinical specimens from patients in five major teaching hospitals in 2020 and found that ESBL was found in 11.9% of E. coli (Hsueh et al., 2001). Further study by Shu et al. showed the prevalence of ESBL increased from 4.8 to 10.0% during seven-year surveillance in a large teaching hospital in north Taiwan (Shu et al., 2010). Further analysis of the distribution of ESBL-producing isolates in different ICUs showed that the most significant increase occurred in medical ICUs, with a peak prevalence rate of 35.9% in 2006 for E. coli (Shu et al., 2010). In a respiratory care center, one retrospective analysis in a tertiary care center from January 2001 to December 2002 reported that the ESBL phenotype was found in 31.4% of E. coli (Lee C. M. et al., 2009). In the respiratory care ward involving patients who required prolonged or long-term mechanical ventilation, Lin et al. reported that the prevalence of ESBL-producing isolates of E. coli was 39.5% (Lin et al., 2013). For nursing home residents, a retrospective study conducted in medical wards of a district hospital in southern Taiwan between July 2009 and June 2011 reported that 52.7–69.5% of E. coli isolates had an ESBL (Liu et al., 2016). Due to chronic exposure to multiple antibiotics, E. coli isolates from a nursing home, respiratory care center, or respiratory care ward seemed to have higher ESBL rates than those from an acute ICU setting.

Among the 404 clinical specimens from patients with community-onset E. coli bacteremia in a medical center in southern Taiwan, Hsieh et al. identified only 19 (4.7%) isolates were ESBL producers (Hsieh et al., 2010). For premature babies with E. coli bacteremia, a small study including 27 cases showed five (18.5%) isolates were ESBL producers (Chen I. L. et al., 2017). Additionally, this study found that the level of serum alanine aminotransferase was significantly lower in the ESBL-producing E. coli group than that in the non-ESBL-producing E. coli group (Chen I. L. et al., 2017).

Regarding the genotype of E. coli causing bacteremia, one study in northern Taiwan including 60 patients with ESBL-producing E. coli bacteremia detected 41 (68.3%) isolates with CTX-M β-lactamases. CTX-M-14 accounted for 31 (75.6%) and CTX-M-3 for 9 (22.0%) of the 41 CTX-M isolates, which were associated with chronic renal failure and ICU stay (Wu et al., 2011). In contrast, the study in southern Taiwan found CTX-M group 9 was the most common genotype causing E. coli bacteremia, especially in pediatric patients (85.7%), and there were more E. coli ST131 in ESBL isolates than in non-ESBL isolates (Tsai et al., 2018). Finally, one retrospective study enrolled 121 adults from southern Taiwan with ESBL-producing E. coli bloodstream infections to investigate their sequence type and virulence factors and showed that positivity for the virulence genes iha, hlyD, sat, iutA, fyuA, malX, ompT, and traT was associated with ST131 positivity (all p < 0.05; Hung et al., 2019). Moreover, iroN positivity was associated with 30-day mortality among bacteremia patients without UTIs (Hung et al., 2019).

E. coli was the most common pathogen causing UTIs. For young patients (0–18 years) with UTI, a 10-year study reported that the ESBL rate had increased from 2% in 2003 to 11% in 2012 (Chen et al., 2014). Another study involving young children aged from 1 day to 36 months reported that the ESBL rate was 3.3% (n = 13) among 421 E. coli isolates and it accounted for 93.3% of ESBL-producing Enterobacterales (Wu et al., 2016). In contrast, another study reported that the prevalence of UTIs due to ESBL-producing E. coli in hospitalized children with community-onset UTIs only increased slightly from 0.59% in 2002 to 0.96% in 2006, however, children with ESBL-producing E. coli UTIs had a longer hospital stay (p = 0.031) than those without (Fan et al., 2014). But a retrospective of 159 infant patients with UTIs found that most of them had no prior history of illness, and no significant risk factors for acquiring ESBL-producing E. coli, such as prior antimicrobial use, hospitalization, UTI, and underlying renal diseases (Cheng et al., 2016). For adult patients, a prospective study included 136 patients with gram-negative bacilli causing community-onset UTIs between August 2009 and January 2012 showed that nine (8.0%) of 111 E. coli isolates had ESBL production (Wu et al., 2014a).

Intra-abdominal infection was another common type of E. coli infection. Surveillance of the pathogens isolated from patients with complicated intra-abdominal infections at five tertiary-care hospitals in Taiwan during the period 2006 to 2010 showed that ESBL production was detected in 7.6% (71/935) of E. coli isolates, and the rate of ESBL production in E. coli was steady (Lee et al., 2012). Moreover, the rate of ESBL production among E. coli isolates collected from patients with nosocomial IAI was higher than those from community-acquired IAI (14.6 vs. 4.7%).

ESBL of ST131 has emerged in many parts of the world since the 2000s, including the US, UK, France, Japan, Canada, India, Kuwait, France, Switzerland, Portugal, and Spain (Coque et al., 2008; Komatsu et al., 2018; Day et al., 2019; Broussier et al., 2020; Duffy et al., 2022). There is no exception for Taiwan. The estimated overall ST131 prevalence in this community UTI cohort increased from 11.2% (in 2002–2004) to 17.4% in 2014–2016 (p < 0.01). Especially ST131 with ESBL phenotype (cefotaxime-resistant group) increased from 33.3% in 2002–2004 to 72.1% in 2014–2016 (p < 0.01; Wang et al., 2021).

In some countries, including the US and Korea, the multidrug-resistant ST131 clone producing CTX-M-15 has emerged as a major clone in both the community and hospital (Peirano et al., 2010; Park et al., 2012; Cho et al., 2015). However, the study about the association between ST131 and CTX-M-15 in Taiwan is scarce. A study of 122 patients with ESBL-producing E. coli bacteremia during a 6-year period from 2005 to 2010 in southern Taiwan reported that the most common ST was ST131 (29.5%, n = 36). CTX-M-producing isolates were identified in 72.2% (26/36) ST131 isolates, including CTX-M-14 (n = 13), CTX-M-15 (n = 9) and CTX-M-3 (n = 4; Chung et al., 2012). In addition, CTX-M has been the most common ESBL in E. coli from Taiwan and in UTI causing ST131 E. coli (Wang et al., 2021). CTX-M-type ESBL gene was detected in 83.5% (66/79) of the ST131 isolates. Among them, ST131 isolates carrying group 1 CTX-M ESBL have emerged in 2014–2016, especially in 2016 (Wang et al., 2021). Since CTX-M-15 belongs to group 1 CTX-M, it is possible that CTX-M-15 might be associated with Taiwanese ST131 E. coli clinical isolates, albeit with no further precise identification in the report (Wang et al., 2021).

One study using the database of healthy inhabitants attending health examinations at a medical center in southern Taiwan in 2017 reported that the prevalence rate of asymptomatic ESBL-producing E. coli fecal carriage in adults was 1.9% (14/724; Wu et al., 2019). In this study, ST131 was found in 22 (3.0%) adults, and underlying cancer and stroke were associated with ST131 E. coli fecal carriage (Wu et al., 2019). Although the ST131 E. coli fecal carriage rate was low among asymptomatic adults in this study (Wu et al., 2019), the prevalence of ST131 among ESBL-producing E. coli was much higher in the following clinical entities. Wu et al. conducted a retrospective analysis of 371 consecutive community-onset non-ESBL producing E. coli bloodstream infections in a 1,200-bed hospital in southern Taiwan in 2010 and found that 60 (16.2%) patients belonged to the ST131 group and clonal group (O25b-ST131) accounted for 5.9% of total isolates (Wu et al., 2014b). In the same institution, Chung et al. showed that ST131 remained the major sequence type, which accounted for 29.5% (36/122) among adult patients with ESBL-producing E. coli bacteremia (Chung et al., 2012). In this retrospective study from 2005 to 2010, clone ST131 were more likely to have secondary bacteremia (OR, 5.05; 95% CI, 1.08 to 23.56) and non-use of the urinary catheter (OR, 3.77; 95% CI, 1.17 to 12.18; Chung et al., 2012). Additionally, another study involving 843 adults presenting with community-onset monomicrobial E. coli bacteremia at a medical center between 2008 and 2013 reported that more elderly (76.5 vs. 64.0%; p = 0.01) or nursing-home residents (12.7 vs. 3.8%; p < 0.001) were found in 102 adults infected by the ST131 clone in comparison to 741 adults by non-ST131 isolates (Wang et al., 2018). Although Chung et al. reported that the ST131 clone was not associated with higher mortality compared with the non-ST131 clone (Chung et al., 2012), Wang et al. found that the ST131 clone was associated with higher 28-day mortality, particularly in those infected by ESBL producers (Wang et al., 2018).

Among patients with UTI, ST131 E. coli still plays an important role in this clinical entity. A national surveillance analyzed the temporal trend of the ST131 clone among urinary E. coli isolates in the community from 2002 to 2016, in which 2,997 outpatient urine E. coli isolates were included and 542 were selected for detection of ST131 based on ciprofloxacin and/or cefotaxime resistance (Wang et al., 2021). Overall, the estimated ST131 prevalence gradually increased from 11.2% (in 2002–2004), 12.2% (in 2006–2008), 13.6% (in 2010–2012), to 17.4% in 2014–2016 (p < 0.01). In the ciprofloxacin-resistant/cefotaxime-resistant group, ST131 increased from 33.3% in 2002–2004 to 72.1% in 2014–2016 (p < 0.01). Moreover, age (≥65 years) and ciprofloxacin resistance were independently associated with ST131 (Wang et al., 2021).

For children aged 0–18 years, a prospective study included 157 isolates from stool specimens collected within 3 days of hospitalization between 2013 and 2014 and showed that among 157 E. coli isolates, 26 (16.6%) and 13 (8.3%) were O25b and ST131 positive, respectively (Huang et al., 2018). Among 13 ESBL-producing E. coli isolates, five (38.5%) belonged to CTX-M group 9, among which 4 (80%) were CTXM-14 and O25b-ST131 positive (Huang et al., 2018). For infants <1 year, a study included infant patients hospitalized for ESBL-producing E. coli-associated UTI between 2009 and 2012 and found that O25b-ST131 accounted for 65% of the 111 isolates (Cheng et al., 2015). Although E. coli O25b-ST131 isolates were more susceptible to trimethoprim/sulfamethoxazole, there were more resistant to ciprofloxacin (Cheng et al., 2015).

Several anthropogenic factors, including the drinking water process, pork consumption, pets, and household density might be associated with ST131 E. coli (Cheng et al., 2022). Compared with families who live in less crowded houses, participants with pets had a similar trend of higher risks of ESBL-producing E. coli, ST131 E. coli, and extraintestinal pathogenic E. coli fecal carriage among those living in houses accommodating relatively more people (Cheng et al., 2022). In addition to the above clinical settings, Chen et al. investigated the epidemiology of ESBL-producing E. coli from multiple rivers in southern Taiwan (Chen et al., 2016). They found that ESBL-producing E. coli mostly belonged to clonal complexes ST10 and ST58, which were geographically related to chicken farms, however, ESBL-producing E. coli ST131 was not detected among the isolates from river water (Chen et al., 2016).

The ST131/O25b strain is a global zoonotic clone of public health concern, however, the prevalence of ST131 E. coli in animal studies was various. One study including 275 E. coli isolated from piglets with diarrhea in swine farms showed that the occurrence rate of ESBL-producing E. coli was 19.7% (n = 54) in southern Taiwan in 2015 (Lee and Yeh, 2017). The ST10 clonal complexes comprised most of the ESBL-producing E. coli strains and the most detected β-lactamase genes were the CTX-M-15 gene (16 of 54; 29.6%) and CTX-M-55 gene (34 of 54; 63.0%), which belong to the CTX-M-1 group (Lee and Yeh, 2017). Another surveillance screened for 283 E. coli isolates in dogs and cats from 2014 to 2017 and found 65 (23%) E. coli (54 from dogs and 11 from cats) with the ESBL phenotype (Huang Y. H. et al., 2020). Additionally, the CTX-M-1 group and CTX-M-2 group were the most identified ESBL gene groups. CTX-M-55 gene was the main ESBL gene within the CTX-M-1 group, whereas the CTX-M-2 group contained only CTX-M-124. Further multilocus sequence typing indicated that ST457, ST131, and ST648 were the most common sequence types, and eight ST131/O25b isolates were identified (Huang Y. H. et al., 2020). Another study investigated the rectal swab specimens from 299 non-infectious dogs and found that the prevalence of ESBL-producing E. coli was 9.4%, and seven isolates were positive for ST131, in which the most predominant subtypes were FimH41 and FimH22 (Chen et al., 2020). Overall, it is worth noting that the potentially wide-spreading CTX-M-55 has been emerging in E. coli isolates from both humans and animals in Taiwan (Jean and Hsueh, 2017).

Three studies before 2000 had reported the prevalence of ESBL-producing K. pneumoniae isolates. In 1998, Jan et al. reported that the frequency of ESBL-producing K. pneumoniae isolates (according to the disk-diffusion method) had increased markedly in the years from 3.4% in 1993 to 10.3% in 1997 in a single center in northern Taiwan (Jan et al., 1998). At the same time, Liu et al. showed that 31 (29.8%) of 104 clinical isolates of K. pneumoniae collected over a period of 8 months were found to be ESBL producers in a district teaching hospital in central Taiwan in 1997 (Liu et al., 1998). In 1999, a multicenter study showed that ESBL production in Klebsiella spp. was found to be 21.7% from isolates in six medical centers in Taiwan (Biedenbach et al., 1999).

Between 2000 and 2010, many studies in different parts of Taiwan or settings showed the various prevalence of ESBL-producing K. pneumoniae. A study in a district hospital in northern Taiwan found that 59 (14%) of 422 isolates of K. pneumoniae collected in 2001 were ESBL-producing strains (Lin et al., 2003). In addition, they demonstrated that tracheostomy and ceftazidime use were risk factors in the acquisition of K. pneumoniae with ESBLs (Lin et al., 2003). Moreover, a significantly rising prevalence of ESBL production among K. pneumoniae in national surveillance for 3 months in 2005 was noted when compared with a previous Taiwanese survey in 2000 (p = 0.002; Jean et al., 2009). Additionally, clonal dissemination (both interhospital and intrahospital dissemination) of ESBL-producing isolates of K. pneumoniae occurred in several regions of Taiwan, mainly in the hospitals located in the northern and central regions of Taiwan (Yu et al., 2004b).

For critically ill patients, a multicenter study in 2000 showed that the percentage of ESBL-producing K. pneumoniae was 11.3% of isolates from various clinical specimens from patients in ICUs (Hsueh et al., 2001). For pediatrics requiring ICU admission, a single medical center in central Taiwan showed that ESBL-producing K. pneumoniae accounted for 20% of K. pneumoniae isolates since 2005 (Lee C. Y. et al., 2009). In a respiratory care center, a retrospective analysis from January 2001 to December 2002 revealed the prevalence of ESBL phenotype was as high as 69.1% of 53 K. pneumoniae isolates (Lee C. M. et al., 2009). For patients with community-acquired liver abscess, a small study showed that seven of 30 K. pneumoniae strains associated with primary liver abscess were found to be ESBL-producers, in which SHV-5a was found in 5, whereas SHV-5 and CTX-M-9 group were detected in 1 strain (Lin et al., 2006). For diabetic patients, a retrospective analysis involving 193 adult diabetic patients with K. pneumoniae bacteremia hospitalized between January 2005 and December 2006 showed that the rate of ESBL infections in the nosocomial group was 11 times higher than that in the community group (45.7 vs. 4.1%, p < 0.001; Tsai et al., 2010). Moreover, ESBL infection accounted for 53% of mortality in the nosocomial group (Tsai et al., 2010).

One study including 57 fistula or graft- or catheter-related ESBL-producing K. pneumoniae bacteremia in patients on maintenance hemodialysis (HD) found most of the patients were elderly, malnourished, and with a history of severe illnesses, broad-spectrum antibiotic use before the onset of bacteremia, and severe septicemia (Yang et al., 2012). Further multivariate analyses revealed that flomoxef use (OR, 3.52; 95% CI, 1.19–58.17), Pitt bacteremia score (OR, 2.92; 95% CI, 1.36–6.26), and catheter-dependent HD >30 days (OR, 5.73; 95% CI, 1.21–63.2) were independently associated with increased mortality (Yang et al., 2012). A longitudinal study involving 1,016 K. pneumoniae isolates from the community- outpatients or those visiting emergency rooms collected during 2002–2012 from the Taiwan Surveillance of Antimicrobial Resistance program showed that the prevalence of ESBL-producers significantly increased from 4.8% in 2002 to 11.9% in 2012 (p = 0.012; Lin et al., 2016).

In 1998, Liu et al. first reported the molecular epidemiology of ESBL-producing K. pneumoniae clinical isolates in Taiwan and found that 22 (71%) of 31 isolates were found to produce SHV-5 (Liu et al., 1998). Additional analytical isoelectric focusing also found seven isolates that produced two unknown ESBLs (with pIs of 7.9 and 7.75), and the pI of 7.9 was later presumed to be CTXM-14 (Liu et al., 1998). In 2000, Yan et al. detected 20 (8.5%) ESBL producers among 234 nonrepetitive clinical isolates of K. pneumoniae, in which a predominance of SHV-12 in 10 strains followed by SHV-5 in 4 strains (Yan et al., 2000). In 2001, Chang et al. repdqorted the stepwise mutations initiated from SHV-1 or SHV-11 into SHV-2, SHV-5, and SHV-12 comprise the evolutionary change responsible for ESBL production based on the analysis of a total of 113 blood culture isolates of K. pneumoniae from 10 hospitals in northern Taiwan (Chang et al., 2001).

In 2002, a multicenter study first identified another rapidly evolving group of CTX-M family enzymes (CTX-M-3 and CTX-M-14) with inter-and intrahospital clonal dissemination in Taiwan (Yu et al., 2002). In 2004, CTX-M-15, an Asp-240-Gly variant of CTX-M-3 with increased catalytic efficiency against ceftazidime, was first described in two of 211 ESBL-producing K. pneumoniae isolates (Yu et al., 2004a). In 2006, Yu et al. found that the insertion sequence IS26 and IS5 were found downstream from the SHV-5 gene in a transferable plasmid pKP53 (Yu et al., 2006a). In 2009, nationwide surveillance from seven medical centers and 13 regional hospitals showed that 102 (43.4%) of 235 ESBL-producing K. pneumoniae isolates were resistant to amikacin (Ma et al., 2009). Moreover, 92 of these 102 (90.2%) isolates were carrying CTX-M-type β-lactamases individually or concomitantly with SHV-type or CMY-2 β-lactamases and CTX-M-type β-lactamase genes belonging to either group 1 (CTX-M-3 and CTX-M-15) or group 9 (CTX-M-14) were found in all amikacin-resistant ESBL-producing K. pneumoniae isolates (Ma et al., 2009). Further molecular typing revealed that the amikacin-resistant ESBL-producing K. pneumoniae isolates were epidemiologically unrelated, and suggested that the acquisition of resistance was not through the spread of a resistant clone (Ma et al., 2009). In 2017, Chang et al. included 51 E. coli transconjugant strains with plasmids from ESBL-producing K. pneumoniae from the Taiwan Surveillance of Antimicrobial Resistance III Program and found that all the 51 plasmids carried a CTX-M gene, the majority of which were CTX-M-3 gene [28/51 (54.9%); Chang et al., 2017]. They found that the most common replicon type of plasmids was incompatibility group (Inc)A/C (60.8%), in which all carried CTX-M-3, CTX-M-14, and CTX-M-15 genes, and some also carried SHV-5 and SHV-12 genes (Chang et al., 2017). In this surveillance, greater than 50% of plasmids fell into clusters, and >60% of cluster-classified plasmids were present in clonally unrelated isolates, which should suggest that horizontal transfer of plasmids in the spread of ESBL genes (Chang et al., 2017).