Clinical and Economic Impact of Third-Generation Cephalosporin-Resistant Infection or Colonization Caused by Escherichia coli and Klebsiella pneumoniae: A Multicenter Study in China

Quantifying economic and clinical outcomes for interventions could help to reduce third-generation cephalosporin resistance and Escherichia coli or Klebsiella pneumoniae. We aimed to compare the differences in clinical and economic burden between third-generation cephalosporin-resistant E. coli (3GCREC) and third-generation cephalosporin-susceptible E. coli (3GCSEC) cases, and between third-generation cephalosporin-resistant K. pneumoniae (3GCRKP) and third-generation cephalosporin-susceptible K. pneumoniae (3GCSKP) cases. A retrospective and multicenter study was conducted. We collected data from electronic medical records for patients who had clinical samples positive for E. coli or K. pneumoniae isolates during 2013 and 2015. Propensity score matching (PSM) was conducted to minimize the impact of potential confounding variables, including age, sex, insurance, number of diagnoses, Charlson comorbidity index, admission to intensive care unit, surgery, and comorbidities. We also repeated the PSM including length of stay (LOS) before culture. The main indicators included economic costs, LOS and hospital mortality. The proportions of 3GCREC and 3GCRKP in the sampled hospitals were 44.3% and 32.5%, respectively. In the two PSM methods, 1804 pairs and 1521 pairs were generated, and 1815 pairs and 1617 pairs were obtained, respectively. Compared with susceptible cases, those with 3GCREC and 3GCRKP were associated with significantly increased total hospital cost and excess LOS. Inpatients with 3GCRKP were significantly associated with higher hospital mortality compared with 3GCSKP cases, however, there was no significant difference between 3GCREC and 3GCSEC cases. Cost reduction and outcome improvement could be achieved through a preventative approach in terms of both antimicrobial stewardship and preventing the transmission of organisms.


Introduction
Escherichia coli and Klebsiella pneumoniae, both species of the family Enterobacteriaceae, are the most prevalent gram-negative bacteria causing intra-abdominal infection, urinary tract infection, and bloodstream infection [1,2], and can be resistant to the widely used antibiotics, such as third-generation cephalosporins, namely third-generation cephalosporin-resistant E. coli (3GCREC) and third-generation cephalosporin-resistant K. pneumoniae (3GCRKP) [3,4]. The World Health Organization (WHO) classified 3GCREC and 3GCRKP as critical-priority bacteria [5]. Alvarez-Uria et al. (2018) pointed out that global resistant prevalence was 64.5% for 3GCREC and 66.9% for 3GCRKP by 2030 [6]. The China Antimicrobial Resistance Surveillance System reported that the average proportion of 3GCREC and 3GCRKP in 2019 was 51.9% and 31.9%, respectively [7], which was higher than the levels in United Kingdom (11.0% and 13.0%) and in Sweden (8.3% and 5.5%) [8].
Third-generation cephalosporin resistance in E. coli or K. pneumoniae is a global concern [9,10]. Infections caused by 3GCREC and 3GCRKP were associated with higher mortality, longer length of stay (LOS), and more economic costs compared with susceptible cases [11][12][13]. de Kraker et al. (2011) showed that 15,183 episodes of 3GCREC were associated with 2712 excess deaths, 120,065 extra LOS, and €18.1 million increased costs in 31 European countries [13]. It was concluded that patients with third-generation cephalosporin-resistant Enterobacteriaceae contributed to 16.1% of hospital mortality, 4.9 days of LOS, and €320 of infection cost in one study by Stewardson et al. (2016) [14]. In addition, colonization of E. coli and K. pneumoniae, as the reservoir for infection with these organisms, was also a risk factor for higher mortality, longer LOS, and increased hospital costs [15,16].
Quantifying clinical and economic outcomes would facilitate strategies towards the containment of third-generation cephalosporin resistance and E. coli or K. pneumoniae. Resistance to third-generation cephalosporins by E. coli or K. pneumoniae, which represented the major mechanism of antimicrobial resistance, had been reported as independently associated with a poor outcome and increased use of healthcare resources [12,17]. However, no significant difference in hospital mortality between 3GCREC and third-generation cephalosporin-susceptible E. coli (3GCSEC) was reported [13,18]. In China, there was only one study exploring longer LOS and higher hospital costs attributable to extended spectrum beta-lactamase (ESBL)-positive intra-abdominal infection caused by E. coli or K. pneumoniae [19]. The clinical and economic outcomes of 3GCREC and 3GCRKP remained largely uninvestigated in China. In this study, we aimed to compare the clinical and economic difference between 3GCREC and 3GCSEC, and between 3GCRKP and third-generation cephalosporin-susceptible K. pneumoniae (3GCSKP), in China.

Study Site
We conducted this study in four tertiary hospitals in China; three in Zhejiang Province (Site 1, Site 3, and Site 4) are a general provincial hospital, general county hospital, and combined traditional Chinese and Western medicine provincial hospital, respectively, and one in Shandong Province (Site 2) is a general provincial hospital. There are 3200, 3500, 1727, 2100 of hospital beds and 170,000, 160,000, 80,000, 50,000 inpatients per year in these four hospitals, respectively.

Study Design and Patients
A retrospective and multicenter study was conducted. We collected data from electronic medical records (EMR) for patients who had clinical samples positive for E. coli or K. pneumoniae isolates, that were detected in any specimens (e.g., blood, stool, cervical, and urethral sources) between 2013 and 2015 [20]. Patients were defined as 3GCREC/3GCRKP cases if patients infected or colonized by E. coli or K. pneumoniae were resistant or intermediate to any third-generation cephalosporin or as 3GCSEC/3GCSKP cases if they were susceptible to all third-generation cephalosporins according to the Clinical and Laboratory Standards Institute (CLSI) definitions [15,21]. We only included the first episode for each patient to avoid duplication. The study was approved by the institutional review board of Zhejiang University School of Public Health, who waived the need for informed consent. All inpatients data were anonymized prior to analysis.

Data Collection
We collected patient characteristics from EMR. The data for each patient included demographics (age, sex, and insurance), comorbidities (disease diagnosis, and Charlson comorbidity index (CCI), hospital events (admitting service, surgical services, and date of hospital and intensive care unit (ICU) admission or discharge), microbiological data, clinical outcomes (discharged alive or death during hospitalization), and economic costs.

Propensity Score Matching
To minimize the impact of potential confounding variables, we performed propensity score matching (PSM) with 1:1 nearest-neighbor matching. PSM, widely used to control for confounding in observational studies, is a powerful statistical matching technique for reducing a set of confounding variables to a single propensity score in order to effectively control for all observed confounding bias [22]. There were two step-by-step rounds of PSM. First, we employed a logistic regression model with third-generation cephalosporin-resistant or-susceptible as dependent variables, and with age, sex, insurance, number of diagnoses, CCI, admission to ICU, surgery, and comorbidities as independent variables. Second, because LOS is the major contributor to additional economic cost, we repeated the PSM including LOS before culture as a potential confounding variable. The generated pairs matched with potential confounding variables were subjected to further analyses of economic costs, LOS and hospital mortality.

Indicators and Statistical Analyses
The main indicators included economic costs, LOS and hospital mortality. The economic costs comprised total hospital cost, medication cost (antibiotic cost), diagnostic cost, treatment cost, material cost, and other costs, and they covered out-of-pocket payment by patients themselves and payments by health insurers. All economic costs were presented in 2015 United States (US) dollars values according to purchasing power parities and the consumer price index of China [23,24].
The Wilcoxon rank-sum test and χ 2 test were conducted to compare the main indicators between 3GCREC and 3GCSEC and between 3GCRKP and 3GCSKP for the quantitative and qualitative variables, respectively. Statistical analyses were performed using STATA. All p-values were two-tailed, and those less than 0.05 were considered statistically significant.
After PSM for potential confounding variables including LOS before culture, the differences in economic costs, LOS and hospital mortality for inpatients with 3GCREC and 3GCRKP were lower than the results after PSM for variables excluding LOS before culture. The differences in total hospital cost, antibiotic cost, medication cost, diagnostic cost, treatment cost, and material cost between the 3GCREC and 3GCSEC groups and between the 3GCRKP and 3GCSKP groups were statistically significant, with median differences of $1140 ($942-$1227), $127 ($127-$147), $515 ($456-$592), $67 ($61-$85), $271 ($245-$296), and $107 ($101-$114), respectively, for inpatients with 3GCREC (Table 2) Table 3). The LOS of inpatients with 3GCREC or 3GCRKP was significantly longer than that of inpatients with 3GCSEC or 3GCSKP, with a median difference of 2.5 days and 7 days, respectively (Table 4). In addition, no significant difference in hospital mortality between the 3GCREC and 3GCSEC groups was found (p = 0.508), but significant difference existed between the 3GCRKP and 3GCSKP groups (p = 0.001) ( Table 5).

Discussion
Previous studies mainly focused on antibiotic utilization and resistance mechanisms and the clinical and economic outcomes of 3GCREC and 3GCRKP in China remained largely uninvestigated. To the best of our knowledge, this is the first study to quantify the clinical and economic outcome of 3GCREC and 3GCRKP in mainland China using the PSM method with large sample size and multiple hospital settings. We focused on E. coli or K. pneumoniae, avoiding non-specific effects from a combination of bacteria [14,25]. In this study, we found that compared with third-generation cephalosporin-susceptible cases, those with 3GCREC and 3GCRKP were associated with significantly increased total hospital cost and excess LOS. In addition, inpatients with 3GCRKP were significantly associated with higher hospital mortality compared with 3GCSKP cases, however, there was no significant difference between the 3GCREC and 3GCSEC groups.
Conducting economic and clinical evaluation for interventions could help to reduce the transmission of 3GCREC or 3GCRKP in hospital settings [26]. It was demonstrated that third-generation cephalosporin resistance increased the economic costs and prolonged the LOS among inpatients with E. coli and K. pneumoniae [11][12][13][14][15]18,19,25,[27][28][29][30][31][32]. For example, Hu et al. (2010) showed that ESBL-positive intra-abdominal infection led to attributable hospital costs and excess hospital stay in China [19]. MacVane et al. (2018) reported that urinary tract infection caused by ESBL-producing E. coli or K. pneumoniae was associated with significant hospital cost and hospital stay in the United States [31]. Meanwhile, one study explored the possibility that colonization with ESBL producing E. coli was associated with longer LOS and higher hospital costs as well [16].
In addition, inpatients with 3GCRKP were significantly associated with higher hospital mortality compared with those with 3GCSKP, which was consistent with other studies [2,32]. However, there was no significant difference in hospital mortality between 3GCREC and 3GCSEC in our study, which was different compared to other studies conducted in European countries [13,18]. Meanwhile, some studies also found there was no difference in hospital mortality between ESBL-producing E. coli cases and non-ESBL-producing cases [16,33]. Different conclusions might be associated with different study design, sample size, geography, resistant pattern, etc. Therefore, this finding needs to be further explored in the future. In addition, the manners in which the use of beta-lactams might affect prevalence of third-generation cephalosporin resistance remained to be fully elucidated [27].
LOS could increase daily bed cost, and might contribute to more treatment service and diagnostic service, therefore, LOS was the major contributor to economic costs [34]. In this study, we applied the PSM method using two step-by-step rounds [29,30,35]. Although the inclusion of LOS before culture as an independent variable in PSM could attenuate the effect of 3GCREC or 3GCRKP on economic costs, LOS and hospital mortality, the conclusion was unchanged when LOS before culture was excluded between the two groups.
This study is not without limitations. First, due to the retrospective nature of our study, it was difficult to distinguish infection or colonization. It was necessary to explore the burden of 3GCREC and 3GCRKP, either infection or colonization, because colonization was an important reservoir for organisms of infection. Prospective studies among patients with infections need to be conducted in the future. Second, PSM was used to balance potential confounding factors, however, some unmeasured variables might still be there. Third, as we had data from between 2013 and 2015 only, we were able to analyze only data corresponding to this study period. Although the study period did not influence the conclusions, future studies with updated data are warranted.

Conclusions
Third-generation cephalosporin resistance increased economic costs and prolonged LOS among inpatients with E. coli and K. pneumoniae. In addition, inpatients with 3GCRKP were significantly associated with higher hospital mortality compared with 3GCSKP cases, however, there was no significant difference in hospital mortality between the 3GCREC and 3GCSEC groups. Given the clinical and economic burden associated with 3GCREC and 3GCRKP that we have demonstrated, efforts to control the development and spread of third-generation cephalosporin resistance and E. coli and K. pneumoniae should be a priority. Cost reduction and outcome improvement could be achieved through a preventative approach in terms of both antimicrobial stewardship and preventing the transmission of organisms. In addition, proper assessment before the empirical use of third-generation cephalosporins is recommended to mitigate costs.