Social-Economic Burden of Patients with Bloodstream Infections Caused by Extended-Spectrum β-Lactamase-Producing Escherichia Coli

Background: The prevalence of infections with extended-spectrum β-lactamase-producing Escherichia coli (ESBL-EC) is increasing worldwide, but the economic impact of ESBL-EC bloodstream infections (BSI) has not been comprehensively evaluated. Methods: A retrospective cohort including patients hospitalized at a tertiary hospital between January 2013 and December 2016 who conrmed with BSI of ESBL-EC or non-ESBL-EC was set. Clinical data and medical costs were collected by chart review of electronic and paper medical records. The socio-economic burden was evaluated with DALYs. Results: A total of 580 patients with E. coli BSI, comprising 333 patients (57.4%) with ESBL-EC BSI and 247 patients (42.6%) with non-ESBL-EC BSI, were identied. There were no signicant differences in comorbidity and severity of patients between ESBL-EC and non-ESBL-EC BSIs. The median length of stay (LOS) after bacteremia was 12 days for ESBL-EC(interquartile range, 7 to 21), versus 11 days for non-ESBL-EC (interquartile range, 7 to 21) (P =0.38), and appropriate empirical antimicrobial therapy occurred in 87.4% versus 89.9% (P =0.353). The mortalities were 20.1% versus 17.4% (P =0.41). Patients with ESBL-EC did not have signicant difference in-hospital medical costs than those with non-ESBL-EC (median, $8048.68 vs $7476.84, respectively,with a difference of $571.84,P=0.321). In non-ESBL-EC group, 247 patients lost 531.05 DALYs in total, with an average of 2.15 DALYs per person,while in ESBL-EC group, 333 patients lost 692.64 DALYs in total, with an average of 2.08 DALYs per person. There is no signicant difference in average DALY(P=0.343). Conclusions: In conclusion, patients with BSI due to ESBL-EC did not cost more than patients with BSI due to dened as a positive blood culture greater than to 48 drugs in culture-negative antimicrobial microorganisms in vitro


Introduction
Escherichia coli, a member of the Enterobacteriaceae, is a main pathogen responsible for community and nosocomial infections, and is the leading cause of Gram-negative bloodstream infections (BSIs) 1 . β-lactamase enzymes production is the common resistance mechanism to β-lactam antibiotics in Gram-negative bacteria. In addition, the production of Extendedspectrum β-lactamase (ESBL) enzymes in bacteria has confered resistance to other types of non-penicillin antibiotics, including uoroquinolones, aminoglycosides, trimethoprim-sulfamethoxazole 2 . Thus, ESBL-producing organisms commonly exhibits a multidrug resistance phenotype. Infections due to β-lactamase-producing E. coli (ESBL-EC) have dramatically increased worldwide, presenting a great public concren. The prevalence of infections with ESBL-producing pathogens has steadily risen since 2000 3 . Lately, a study showed that a high prevalence of community-acquired ESBL-producing Enterobacteriaceae infections (46.5%, 256/550) had been reported in public county hospitals in China 4 .Consistent herewith, the presence of faecal ESBL-producing Enterobacteriaceae from healthy individuals are also very high in China ranging from 42.0% to 82.6% 5; 6; 7 .
Although the problem of ESBL resistance has attracted great attention of the public, the magnitude of the impact of drugresistant bacteria on clinial and economic outcomes remains largely unknown. Therefore, we aimed to quantify the potential clinial and economic impact of ESBL production. In some studies, they showed increased mortality associated with ESBLpositive infection 8; 9 ,while our previous study 10 had come to the opposite conclusion that patients with BSI due to ESBL-EC did not show a higher mortality and a longer hospitalization than patients with BSI due to non-ESBL-EC.
Furthermore, there are few assesments of speci c economic impact of ESBL production on patient outcomes 8; 9; 11 . These articles mainly studied the direct economic burden of drug-resistant bacterial infections, and did not pay attention to the indirect economic burden caused by drug-resistant bacteria infections. Disability-adjusted life years (DALYs) is a widely used metric for estimating disease burden, which developed and used by experts from Harvard University School of Public Health and the World Health Organization in 1993. The Global Burden of Disease (GBD) study systematically compared the magnitude of health losses caused by different diseases worldwide. DALYs was suscessfully used by GBD to measure quantitatively health losses. It is a summary measure that combines the time lost due to premature mortality, expressed as years of life lost (YLL), with the time lived in states worse than full health, expressed as years lived with disability (YLD) 12 . One DALY can be thought of as one year of 'healthy' life lost due to different diseases. The sum of these DALYs can be thought of the gap between current health status and that of an ideal health situation, that is normative reference population lives to an advance age, free of disease and disability 13 .
To analyze the socio-economic impact of ESBL-EC BSI, we conducted a retrospective cohort study to compare direct costs and indirect costs between inpatients with ESBL-EC and those with non-ESBL-EC.

Patients
A retrospective cohort study of adult inpatients with E. coli bacteremia at a 2500 bed teaching hospital(Zhejiang, China) from January 2013 through December 2016 was conducted.Inclusion criteria were as follows: (1) all patients had a positive blood culture for E. coli; (2) clinical manifestations of bloodstream infection; and (3) hospitalization with complete clinical microbiological and cost data for analysis. Patients were excluded if they had incomplete medical records or their age younger than16 years. Only data from the rst episode of bacteremia of inpatients experiencing more than one episode of E. coli bacteremia within a 6-month period were included. The patients infected with ESBL-EC were referred as "cases" and patients infected with non-ESBL-EC were refrred as "controls" in the study.

Microbiological tests
These isolates were identi ed by the Vitek2 system (bioMérieux, Marcy-l' Etoile, France) and antimicrobial susceptibility of these strains were assessed. According to the guidelines of the Clinical and Laboratory Standards Institute (CLSI) standards (2018) 14 , ESBL production was determined by thedouble-disk test using combination of ceftazidime and ceftazidimeclavulanic acid or cefotaxime and cefotaxime-clavulanic acid.

De nition
Hospital-acquired bacteremia was de ned as a positive blood culture obtained greater than or equal to 48 h after admission. Community-acquired bacteremia was de ned as a positive blood culture taken on admission or less than 48 h afterr admission. Severity of illness at the time of BSI onset was assessed by the Acute Physiology and Chronic Health Evaluation (APACHE) II scores and Pitt scores 15 . The Charlson comorbidity index was calculated to determine overall systemic health status of patients 16 . Initial antibiotic treatment was de ned as the drugs administered empirically in culture-negative situations. The initial antimicrobial therapy was considered effective if the empirical administration of at least one antibiotics being active against isolated microorganisms in vitro susceptibility testing 17 .

Data collection
For all patients enrolled in this study, clinical and laboratory data were collected from electronic medical records, including: patient demographics (sex; age), comorbid illnesses, Charlson comorbidity index, day of BSI onset, intensive care unity (ICU) stay before bacteremia, organ transplantation during hospitalization, severity of illness (APACHEII score and Pitt score), overall hospital length of stay (LOS) and LOS before/after the infection; microbiological data, susceptibility of empirical antimicrobials and targeted antimicrobial, and clinical outcome recorded as"death" or "discharged alive". The study complies with the Declaration of Helsinki; Ethics Committee approval was received from the Hospital Ethics Committee.

Cost analysis
The cost was obtained from the hospital information system. The patients direct costs and indirect costs (i.e. productivity losses due to absenteeism and mortality) were analyzed. All the costs were converted into US dollars($) with an exchange rate (average: $1 = 6.33 Renminbi).

Direct cost
The total direct costs comprised of room and board, nursing, medicines(including antibacterial agents, traditional chinese medicines), oxygen inhalation, mechanical ventilation, blood transfusion, operation, laboratory tests and images.

Indirect economic burden
The indirect economic burden of E. coli bacteremia was analyzed through DALYs and human capital methods, which equal to DALYs multiplied by Gross Domestic Product (GDP) per capita multiplied by productivity weight. And the GDP per capita in DALYs for E.coli BSI are calculated by summing the YLLs for all deaths caused by this disease and the YLDs for people living in states of less than good health caused by this disease 13 . These factors, such as life expectancy, age, future time and disability, were included in the YLLs and YLDs calculation.
The formula for YLLs is described below. The calculation YLLs [γ,K,β] is used to signify key factors (age weight and discount rate). Values were recommended and used by Murray and Lopez 20; 21 ie. γ = 0.03, K = 1 and β = 0.04. K = age weighting modulation factor; C = constant; γ = discount rate; a = age of onset of disability; β = parameter from the age weighting function; L = standard expectation of life at age a.
The difference with the formula for YLLs is that D(the disability weight) was added in the formula for YLDs [γ,K,β], as follows: K=age weighting modulation factor; C=constant; γ = discount rate; a=age of onset of disability; β = parameter from the age weighting function; L= duration of disability; D=disability weight.
This formula uses the values recommended by the World Health Organization (WHO) 22 , constant valve is 0.1658 22 . The value of disability weight (D) ranges from 0 to 1 according to the Global Burden of Disease (GBD) template provided by the WHO 12; 23; 24 . Since the recent GBD studies did not mention the standard for BSI, we decided that the evaluation of D is based on the acute infection, which is divided into 0.006, 0.051 and 0.133 according to severity of the diseases in mild, moderate and severe conditons 12 . In order to calculate L( the year lost by death and discounted by disability),We used "standard expected years of life lost" (SEYLL) as a good approximation of life expectancy 22 .

Statistical methods
Date were expressed as mean standard deviation (S.D.) or median (interquartile range [IQR]) for continuous variables and percentage n(%) for categorical variables. For statistical analysis, inter-group differences were tested using Student's t-test (for variables with normal distribution) or the Mann-Whitney U test (for variables with non-normal distribution) for continuous variables and χ2 or two-tailed Fisher exact test for Categorical variables. P values less than 0.05 were considered statistically signi cant. Statistical analyses were performed using SPSS version 23.0 package (SPSS, Chicago, IL, USA).

Cost analysis Direct costs
The median cost for patients with ESBL-EC was $8048.68 and for patients with non-ESBL-EC was $7476.84, respectively (P 0.05) ( Table 2). The direct costs after E.coli BSI also did not show signi cat difference between the tow groups.
In both groups, the cost of antibiotics accounted for 19.79% and 19.13% of the medicine, respectively. The median cost of antibiotics for patients with ESBL-EC BSI during hospitalization was $1592.50, which was higher than that for patients with non-ESBL BSI (median, $1430.06, P = 0.252). To exclude the effects of antibiotic costs before infection, we also compared the cost of antibiotics after infection between the two groups, no signi cant difference between the two groups was evident. (Table 1). BLBLI and carbapenems were the main antibiotics used in the two groups.

Indirect loss
In  Table 2). Among all paients who with E.coli bloodstream, the heaviest total DALYs and indirect economic loss were in the age group of 45-69, and the least total DALYs and indirect economic loss were in the age group of > 74 years (Fig. 1.) .

Discussions
The information on the economic burden of infections caused by antibiotic-resistant pathogens can only be obtained from observational cohort studies, which are highly susceptible to bias and confoundings 25 . So in the process of assessing the economic impact of the production of ESBL, the rst thing we need to do is to eliminate the effects of intergroup differences on patients' characteristics. Therefore, it is crucial to adjust for the relevant confounders when investigating the link between drug resistance and economic burden. In our study,there were no signi cant differences in baseline characteristics of patients (including patient demographics, comorbidity status, severity of illness, overall hospital length of stay (LOS) and LOS before/after the infection; susceptibility of empirical antimicrobials demographics, mortality) between ESBL-EC and non-ESBL-EC group were found.Therefore, We can directly compare the economic burden of the two groups without considering the confoundings.
Our study found that the production of ESBL did not lead to signi cant increase in direct costs. The direct cost of a ESBL-EC infection was $8048.68 and a non-ESBL-EC infection was $7476.84, with a difference of $571.84. However, Schwaber et al. 9 found that the average medic cost due to the production of ESBL was $9,620. Tumbarello et al. 8 33 . In the short term, the use of second and third line antibiotics can improve the cure rate, reduce the direct and indirect costs for individual and society, but in the long term, the use of antibiotics across the ladder will lead to the increase and spread of resistance of advanced antibiotics, so for the rational application of antibiotics, clinicals need to balance individual and societal needs.  34 . In our study, E.coli bloodstream had a signi cantly lower burden compared to the infctious diseases discussed above (except in uenza). We can also infer from these data that the duration of E.coli bloodstream is not as long as that of pneumococcal disease, hepatitis B, Haemophilus in uenzae disease, etc. For E.coli bloodstream, 0.07 DALYs were produced due to the production of ESBL, which is seven times the burden of in uenza. Hence using DALYs can quantify and compare the burden of different diseases. espically using DALYS to evaluate the burden of antimicrobial resistant pathogens can help decision makers to measure how much resources and energy should be invested to track and control the spread of antimicrobial-resistant organisms.
However, there are still some limitations in our study. At rst, this was a retrospective single-center study with inherent biases.The disease prevelence and treatment options in this hospital might have had an impact on the results. Andthe situation inother medical institutions or healthcare systems might be different. Secondly, this was an observational study, not a randomized controlled trial, so our ngdings are susceptible to unmeasured confounding at both the hospital and patient levels.Thirdly, due to di culties about statistic measurement and calculation, our study ignored direct non-medical costs and intangible economic burden including transportation costs, food costs and others, which may lead the results could not fully represent the socio-economic burden of disease.
In conclusion, patients with BSI due to ESBL-EC did not cost more than patients with BSI due to non-ESBL-EC. This phenomenon may be attributed to timely and effective antibiotic treatment. But the antimicrobial stewardship should be implemented to avoid overaggresive use of second and third line antibiotics in sensitive bacterial infctions.

Notes
Ethics approval and consent to participate: Ethics approval for this study was submitted and approved through Research Ethics Committee of the First A liated Hospital, College of Medicine, Zhejiang University. The consent to participate was waived by our institutional review board since this study was retrospective data collection.

Consent to publish:
Not applicable.
Availability of data and materials: The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

Potential con icts of interest:
None of the authors report con icts of interest.

Declarations
Ethics approval and consent to participate: Ethics approval for this study was submitted and approved through Research Ethics Committee of the First A liated Hospital, College of Medicine, Zhejiang University. The consent to participate was waived by our institutional review board since this study was retrospective data collection.
Consent to publish: Not applicable.
Availability of data and materials: The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.
Potential con icts of interest: None of the authors report con icts of interest.

Page 8/17
Financial support: This work was partially supported by grants from The National Key Research and Development Program of China (2017YFC1200203) and National Natural Science Foundation of China (81971984) .
Authors' contributions: YHX and YW designed the study and created study protocols. TTX, YYZ and JY performed the data collection, data analysis and drafts the manuscript. KY, QXL and PS performed data collection and helped to review the manuscript. All authors read and approved the nal manuscript.