Clinical and Economic Burden of Bacteraemia Due To Multi-Drug-Resistant Organisms in Korea: A Prospectively Selected Case Control Study

Kyoung-Ho Song Seoul National University Bundang Hospital Chung-Jong Kim Ewha Womans University https://orcid.org/0000-0002-9987-6533 Nam-Kyong Choi Ewha Womans University Jeonghoon Ahn Ewha Womans University Pyoeng Gyun Choe Seoul National University Hospital Wan Beom Park Seoul National University Hospital Nam Joong Kim Seoul National University Hospital Hee Jung Choi Ewha Women's University Mokdong Hospital Ji Yun Bae Ewha Women's University Mokdong Hospital Eu Suk Kim Seoul National University Bundang Hospital Hyunju Lee Seoul National University Bundang Hospital Jeong Su Park Seoul National University Bundang Hospital Younghee Jung Hallym University Sacred Heart Hospital Seung Soon Lee Hallym University Sacred Heart Hospital Kyung-Hwa Park Chonnam National University Hospital Sook-In Jung Chonnam National University Hospital Yeon-Sook Kim Chungnam National University Hospital Ji-Hwan Bang Seoul Metropolitan Boramae Hospital: Seoul National University Seoul Metropolitan Government Boramae Medical Center Shinwon Lee Pusan National University Hospital Yu Min Kang Kangwon National University Hospital Yee Gyung Kwak Inje University Ilsan Paik Hospital Hong Bin Kim (  hbkimmd@snu.ac.kr ) Seoul National University Bundang Hospital https://orcid.org/0000-0001-6262-372X


Introduction
Infectious diseases, particularly nosocomial infections, cause tremendous burden, both in the hospital and in society. The burden of antibiotic-resistant infections, in particular multidrug-resistant organisms (MDROs), is more burdensome than that for susceptible pathogens. Therefore, concerns about the economic burden of MDROs are increasing. In 2019, the Centres for Diseases Control and Prevention estimated that, in the United States, more than 2,800,000 people were affected by antibiotic-resistant infections and 35,900 people consequently died. [1] In Spain, the economic burden of carbapenem-resistant Gram-negative bacteria was estimated at 472 million euros. [2] We previously reported the estimated annual incidence of invasive methicillin-resistant Staphylococcus aureus (MRSA) infection at 13,000 cases in 2010 in South Korea. [3] We also evaluated the economic burden of nosocomial MRSA bacteraemia at 60 million dollars yearly, in a nationwide study. [4] However, there is a general lack of studies on the clinical and societal burden of major MDROs.
In South Korea, six MDROs have become nationally-noti able infectious diseases (NNID) since 2010 with documented prevalence at hospitals through a national sentinel surveillance programme by the Korea Disease Control and Prevention Agency (KDCA). The NNIDs include vancomycin-resistant S. aureus (VRSA), methicillin-resistant S. aureus (MRSA), vancomycin-resistant Enterococcus (VRE), multidrug-resistant Pseudomonas aeruginosa (MRPA), multidrug-resistant Acinetobacter baumannii (MRAB), and carbapenem-resistant Enterobacteriaceae (CRE). However, since the KDCA only reported on MDRO incidence when there were blood samples, the severity of the infection and economic burden of MDRO are unknown. [5] Hence, in this study, we evaluated the clinical outcomes of MDRO infections and estimated the medical costs and socioeconomic burden of MDRO bacteraemia, nationally, in Korea.

Study design
We prospectively searched for and collected MDRO bacteraemia cases with matched controls from September 2017 to February 2018 using the multistate model of Stewardson et al.[6] Patients were classi ed into the MDRO (R-group), susceptible organism (S-group), and no (N-group) infection groups. Hospital costs and length of stay (LOS) were compared among the groups.

Setting
The study was performed in 10 secondary or tertiary hospitals selected in South Korea based on the regional distribution of the hospitals (see Supplementary Figure S1).

Participants
We collected bacteraemia cases caused by S. aureus, Enterococcus, A. baumannii, P. aeruginosa, and Enterobacteriaceae. All S. aureus, A. baumannii, and P. aeruginosa bacteraemia cases, regardless of antibiotic susceptibilities. All R-group cases of bacteraemia due to VRE or CRE, selected based on prede ned criteria, were collected. The corresponding susceptible or no-infection controls were selected and matched in a 3~5:1 ratio to MDRO cases, based on the following ve criteria: similar principal diagnosis at the time of admission, similar major surgery or intervention during the hospitalization, age (± 10 years), sex, and ± 60 days of admission date. For patients that experienced bacterial infection during the same hospital course, the control case was discarded, and another was selected. MDRO bacteraemia cases that developed within 48 h of admission or patients whose duration of admission was over 180 days were excluded from the case-control matching.

Variables
Baseline characteristics, including route of admission, length of hospital stay (LOS), and underlying diseases, were documented, as were data on severity of infection (reported as the sequential organ failure assessment [SOFA] score), portal of bacteraemia, duration of post-bacteraemia hospital stay, and mortality. The hospital costs of each patient were also recorded.

Statistical analyses
Estimation of additional medical costs related to MRDO bacteraemia We estimated the total additional direct medical costs associated with each MDRO bacteraemia, calculated by multiplying the extra cost per patient and the annual number of patients per bacteraemia type, using the KDCA surveillance data. [5] The additional cost per case in the R-group was calculated by subtracting the mean hospital cost of the S-group or N-group from that of the corresponding R-group. The currency exchange rate applied in our calculation was 1,110 Korean won to 1 US dollars.
Estimation of the number of unexpected mortalities due to MDRO infection, nationwide Patients in the R-group were categorised according to age groups with -10 year intervals, to calculate the ratio of patients per age group among the total deaths. The estimated number of deaths due to each type of MDRO bacteraemia nationally was calculated by multiplying the number of MDRO bacteraemia cases in the national surveillance system and the 90-day mortality rate of our data. See web-only Supplementary Tables S1 and S2 for the total number of cases per group.
Using the mortality rate per age group, the estimated total number of deaths in 1 year was calculated per age group. We assumed that the age distribution of patients and the mortality rate obtained from the 10 hospitals and those generated nationwide were similar.

Estimation of additional socioeconomic costs of MRDO bacteraemia
The socioeconomic burden included: 1) the cost of caregiving, calculated by multiplying the daily cost of hired caregivers ($59.10 [65,000 Korean won], obtained from the caregiver association and the excess LOS; 2) the productivity loss due to unexpected death, calculated from the number of deaths associated with MDRO bacteraemia and the annual wages reported by the Ministry of Labour in Korea (Labour Statistics of Korea, Ministry of Labour 2017; available from http://wage.go.kr/index.jsp). The productivity loss due to the unexpected death of a given patient was the sum of the annual wages up to the time that patient would have reached 65 years of age, if he or she had not died. The annual discount rate was considered as 5%. Only patients aged < 65 years, the mandatory retirement age for almost all professions, were included in the calculation of productivity loss.

Sensitivity analyses
Three different models were built. We set a basic model and two other modi ed models (each using either the lower or the upper limits of values). In this analysis, we calculated the economic burden by changing the caregiver employment rate, the patient employment rate, and the discount rate. Sensitivity analyses were performed to assess the impact of changing variables.

Ethical review
This study was approved by the Institutional Review Board (IRB) of Seoul National University Bundang Hospital (IRB no. B-1804-463-105) and also the IRBs of each participating hospital. Informed consent was waived by each IRB.
The rate of infection acquired while in an intensive care unit was lowest for MRSA (16.5%) and highest for MRAB (58.6%) bacteraemia. Although the median LOS was longest for CRE (53 days) the median post-bacteraemia LOS was longest for MRPA (33 days). The in-hospital mortality rates, highest for CRE (65.0%), were 62.1% and 49.5% for MRAB and VRE bacteraemia, respectively, and lowest for MRPA (16.7%) and MRSA (28.1%) bacteraemia. The 90day mortality rate was higher in MDRO than in susceptible bacteraemia patients for S. aureus and A. baumannii infection. Baseline characteristics and clinical outcomes are shown in Table 1. See web-only Supplementary Tables S1 and S2 for the mortality rates by age groups. Survival curves are shown in Figure 2.  Table 2, the mean LOS difference was 1.0 day between the R-group and S-group, and the mean difference in hospital cost was $1,089 for S. aureus bacteraemia; these were longest and highest in the R-group for Enterobacteriaceae, respectively. Compared to the S-group, R-group cases caused 2.3 additional days of hospital stay and $7,507 additional hospital costs for A. baumannii bacteraemia.  Table 3, the mean LOS difference between the R-group and N-group was 24 days, and the mean difference in hospital cost was $15,768 for S. aureus bacteraemia; these were longest and highest in the R-group for Enterobacteriaceae, respectively. Compared to the N-group, the R-group was associated with 24 additional days of hospital stay and $35,682 additional hospital costs for A. baumannii bacteraemia.

Total Burden Of Mdro Infection And Sensitivity Analyses
The total socioeconomic burden of MDRO infection, estimated as the total additional medical cost, cost of hiring a caregiver, and the sum of productivity loss due to unexpected death, was $294,505,002 (   Previous studies on the nationwide burden of disease have mainly focused on chronic diseases such as chronic obstructive pulmonary disease, [15] diabetes mellitus, [16,17] and dementia.
[18] These studies assessed the changes in quality of life due to diseases. On the contrary, our study focused on acute diseases that are associated with higher mortality and morbidity during relatively short-term follow up periods. In our study, the 90-day disease mortality rates were between 16.7% and > 60%. Therefore, when estimating the economic burden, we focused on the LOS and unexpected mortality, rather than the long-term change in quality of life. Infections with MDRO are mostly nosocomial. Therefore, consumption of additional medical resources, prolonged hospital stays, and deaths are unexpected, unpredictable, and unfavorable, but in many cases are preventable with proper investment of resources.
We estimated nationwide trends based on data from 10 sample hospitals. This naturally led to uncertainty in the results. In South Korea, the status of major MDRO infections has been documented via a national surveillance program since 2010. Hence, we attempted to reduce estimation errors by using a national data. We used data from 10 study hospitals to assess the differences in medical expenses between patients and controls, in the LOS, and mortality rates by age; and subsequently, estimated the nationwide impact based on national surveillance data. Sensitivity analyses were also conducted to improve our results.
This study had some limitations. First, some medical costs, such as productivity loss due to the extension of the hospitalization period, loss of work ability, and others factors such as social burden due to deterioration of quality of life from illnesses, were not considered in the analyses. Although these are important components of indirect medical costs, it is di cult to calculate these costs with only the additional information obtained through socioeconomic status and patient questionnaires. In addition, in patients with acute infections, these indirect medical expenses accounted for a relatively small portion of the total expenses. Second, we used a multistate model with a case-case-control design. We selected control with susceptible organism infection and those without infection but with similar major clinical factors of MDRO infection cases, to calculate the differences in the LOS and cost, between the two groups. However, in the process of selecting possible similar control groups of patients, some patients were not selected in the control groups and were excluded from analyses.

Conclusions
In conclusion, ve types of MDRO bacteraemia infected 7,979 patients, with 3,280 deaths, and caused $294,505,002 of socioeconomic burden, ranging from $170,627,020 to $416,094,679 dollars. Proper investigation and investment of infection control measures should be introduced to immediately decrease the burden of MDRO infections.