Frequent Prescription of Antibiotics and High Burden of Antibiotic Resistance among Deceased Patients in General Medical Wards of Acute Care Hospitals in Korea

Background Antibiotics are often administered to terminally ill patients until death, and antibiotic use contributes to the emergence of multidrug-resistant organisms (MDROs). We investigated antibiotic use and the isolation of MDROs among patients who died in general medical wards. Methods All adult patients who died in the general internal medicine wards at four acute care hospitals between January and June 2013 were enrolled. For comparison with these deceased patients, the same number of surviving, discharged patients was selected from the same divisions of internal medicine subspecialties during the same period. Results During the study period, 303 deceased patients were enrolled; among them, 265 (87.5%) had do-not-resuscitate (DNR) orders in their medical records. Antibiotic use was more common in patients who died than in those who survived (87.5% vs. 65.7%, P<0.001). Among deceased patients with DNR orders, antibiotic use was continued in 59.6% of patients after obtaining their DNR orders. Deceased patients received more antibiotic therapy courses (two [interquartile range (IQR) 1–3] vs. one [IQR 0–2], P<0.001). Antibiotics were used for longer durations in deceased patients than in surviving patients (13 [IQR 5–23] vs. seven days [IQR 0–18], P<0.001). MDROs were also more common in deceased patients than in surviving patients (25.7% vs. 10.6%, P<0.001). Conclusions Patients who died in the general medical wards of acute care hospitals were exposed to more antibiotics than patients who survived. In particular, antibiotic prescription was common even after obtaining DNR orders in patients who died. The isolation of MDROs during the hospital stay was more common in these patients who died. Strategies for judicious antibiotic use and appropriate infection control should be applied to these patient populations.


Introduction
Antibiotic use contributes to increased antibiotic resistance due to the selection and expression of antibiotic resistance genes in bacterial populations [1]. Antibiotic abuse can result in antibiotic resistance in individual patients. Antibiotic resistance has been associated with increased attributable length of hospital stay, mortality, and health care costs [2].
Physicians and family members often consider antibiotics to be a minimum treatment requirement for terminally ill patients, and they are often administered until death in those patients [3][4][5]. As in other Asian countries where palliative and hospice care services are not widely utilized for patients with terminal illnesses [6,7], most terminally ill patients in Korea are admitted to acute care hospitals near the end of their lives in order to receive life-sustaining treatments [7].
A recent investigation of intensive care units (ICUs) reported that dying patients without withdraw orders received more antibiotics and developed more multidrug-resistant organisms (MDROs) [8]. Patients who acquire MDROs before death may serve as MDRO reservoirs, transmitting the organisms to surviving patients in the hospital setting [8,9].
While there have been many studies on antibiotic use and resistance in critically ill patients in ICUs and in patients who die in palliative and hospice care settings [3][4][5][10][11][12], scant attention has been paid to antibiotic use in patients who die in the general hospital wards. We hypothesized that a certain proportion of patients who died in the general medical wards of acute care hospitals might be exposed to antibiotics before death, and that the isolation of MDROs might be common in these patients during their hospital stay. The aim of this study was to examine the antibiotic use and isolation of MDROs among patients who died in the general medical wards of acute care hospitals and to compare these characteristics with those of surviving, discharged patients.
Patients who were hospitalized 2 days or >60 days, were transferred to other hospitals, or were discharged against medical advice were excluded.

Data collection
Medical records were reviewed retrospectively, and data were collected using standardized case report forms. The data obtained included demographic characteristics, length of hospital stay, underlying diseases or conditions, and either clinical or microbiological bacterial infection during the admission and at the time of death or discharge. Data on antibiotic exposure for more than 24 hours and isolation of MDROs during the admission were also collected. An antibiotic therapy course was defined as the use of different antibiotics for at least one day during admission. Days of antibiotic use were defined as the total number of days that a single antibiotic (parenteral or oral) was administered, regardless of dosage [13]. MDROs in this study included methicillin-resistant Staphylococcus aureus, vancomycin-resistant enterococci, third-generation cephalosporin-resistant Enterobacteriaceae, and carbapenem-resistant non-fermenters, including Acinetobacter baumannii and Pseudomonas aeruginosa [14]. In addition, a do-notresuscitate (DNR) order was defined as present if the patient had a preprinted DNR form or recorded verbal communication in their medical record [15].

Statistical analysis
Bivariable analysis was performed separately for each variable. P values were calculated using chi-square or Fisher's exact tests for categorical variables. Continuous variables were compared using Student's t (for normally distributed variables) or Mann-Whitney U tests (for non-normally distributed variables). P values <0.05 were considered statistically significant. All variables that were statistically significant in bivariable analyses were included in logistic regression analysis. Results from the multivariable analysis are presented as adjusted odds ratios (AORs) with 95% confidence intervals (CIs).

Results
During the study period, a total of 15,777 patients were discharged from the general internal medicine wards of four hospitals, while 377 died in the hospital. Of these deceased patients, 74 were excluded, and the remaining 303 comprised the study population ( Table 1). The same number of surviving patients was selected from among the 15,400 surviving patients who were discharged. Overall, the most common underlying disease was solid tumor (75.6% vs. 58.7%, P<0.001), followed by cardiovascular disease (42.9% vs. 34.7%, P = 0.037) and diabetes mellitus (27.4% vs. 24.8%).
Of the 303 deceased patients, 265 (87.5%) had DNR orders in their medical records ( Table 2). On average, the first record of DNR order occurred on the 12th day (median, interquartile range [IQR] 4-21) of a 14-day hospital stay (median, IQR 8-26). The median time from DNR order to death was two (IQR 1-6) days, and 53 patients (19.6%) had DNR orders on the day of their death. Among deceased patients with DNR orders, 59.6% continued to receive antibiotics after providing a DNR order. Antibiotics were administered for a median of two (IQR 0-5) days after the day when DNR consent was provided in those patients. Comparison of deceased patients with DNR orders to those without orders revealed no significant differences in the proportion of patients that received antibiotic therapy (87.9% vs. 84.2%, P = 0.599) or the number of days of antibiotic use (median 13 [IQR [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22] vs. 14 [IQR 4-23] days). Malignancy was the most common cause of death (62.4%), and 21 patients (6.9%) received cardiopulmonary resuscitation during admission. Table 3, bacterial infections were more common in patients who died than in those who survived (57.8% vs. 40.6%, P<0.001). Antibiotic therapy lasting more than 24 hours was also more common in deceased than in surviving patients (87.5% vs. 65.7%, P<0.001). The median number of antibiotic therapy courses were two (IQR 1-3) and one (IQR 0-2) for deceased and surviving patients, respectively. Third-generation cephalosporins were the most commonly prescribed antibiotics in both patient groups (43.9% and 39.3%). However, antibiotics were used for significantly longer periods in deceased patients than in those who survived

Discussion
This study revealed extensive antibiotic exposure before death among deceased patients, even those who had provided DNR orders, in the general medical wards of acute care hospitals. Isolation of MDROs was more common in patients who died than in those who survived. Longer antibiotic use duration was associated with dying patients who had MDROs during admission. These findings support our hypothesis but a causality between the antibiotic exposure and isolation of MDROs in the deceased patients could not be defined because of limitations of this study (refer to below). After opioid analgesics, antibiotics are the most commonly administered drugs on the day of death in terminally ill patients, and many of these patients receive antibiotics until death in Korea [5]. In this study, about 90% of patients who died in the general internal medicine wards received at least one course of antibiotic therapy during their hospital stay. Given the proportion of patients with bacterial infection who received antibiotic therapy during admission, antibiotics might be abused at some point during admission in about 30% of deceased patients. It has not been determined whether the use of antibiotics for end-of-life treatment is appropriate and beneficial in terms of improving symptoms in terminally ill patients [7,12]. Furthermore, aggressive treatment of infection has not been found to alter the underlying disease process or improve survival rates among patients with terminal illness [16]. Considering the risk of antibiotic resistance and the paucity of novel antibiotics in the development pipeline, the use of antibiotics in terminally ill patients should be rationalized [16].
Indiscriminate use of antibiotics in terminally ill patients may cause unintended consequences, including individual patient burdens as well as burdens extending to other patients in the same healthcare setting through the possible selection for antibiotic resistance [4]. Although the deceased patients with MDROs had more bacterial infections during admission in this study, multivariable analyses indicated that longer duration of antibiotic use was associated with the isolation of MDROs in these patients. This result is consistent with previous findings that antibiotic courses of longer duration are associated with higher rates of resistance [17]. MDROs that colonize or infect deceased patients can contaminate the environment or the hands of healthcare workers and spread to surviving patients or vice versa [9,11]. Therefore, careful selection of indications for and duration of antibiotic use at the end of life is important to prevent the development and dissemination of antibiotic resistance.
In Asia, physicians are typically more aggressive in their end-of-life treatment compared with their Western European counterparts due to cultural differences [18]. They are less likely to withhold or withdraw life-sustaining treatment, and they will initiate antibiotic treatment in terminally ill patients [18]. In this study, DNR consent was obtained from 87.5% of 303 patients who died in general medical wards in Korea. Among those deceased patients with DNR orders, 59.6% received antibiotic therapy even after providing a DNR order, although some of them had bacterial infections at the time of death. In addition, antibiotics were administered to patients who died regardless of the presence of DNR orders. This is consistent with the view that physicians may feel more comfortable in continuing to try to correct a theoretically reversible condition using antibiotics even in the face of an irreversible dying process [19].
This study has some limitations. First, we investigated the isolation of MDROs based on the results of clinical and hospital surveillance cultures limited to the hospital stay during the study period. Since the microbiology data before this admission and data on the time of isolation for each MDRO were not collected, we could not discriminate whether the MDROs isolated during this admission were already colonized before this admission or newly acquired. As a result, we could not determine the causal relation between the antibiotic use and acquisition of MDROs in the study patients. Second, since we included all patients who died in the internal medical wards, and not just those who were terminally ill, there may be differences in the population of deceased patients compared to previous studies. However, because almost 90% of patients had DNR orders, this difference was likely slight. Third, no was made distinction between colonization and infection. We did not determine the appropriateness of the antibiotic therapy administered to the study patients, so we were unable to estimate the proportion of controlled antibiotic use in patients who died in the general medical wards. Nevertheless, based on the finding in this study that many of the deceased patients had MDROs during their hospital stay before their deaths, it is likely that these patients with MDROs could be sources of transmission to other patients.
In conclusion, our study found that a substantial proportion of patients who died in the general medical wards of acute care hospitals were exposed to antibiotics before death. Antibiotic prescription was common even after obtaining their DNR orders in those deceased patients. In addition, the isolation of MDROs during their hospital stay was more common in patients who died than in those who survived. Because these patients may represent reservoirs of resistant bacteria that could be transmitted to surviving patients in hospital settings [8,9], strategies for judicious antibiotic use and appropriate infection control should be applied to this patient population. Further prospective studies are needed to evaluate whether overuse or abuse of antibiotics in the dying patients in general medical wards contributes to the acquisition and transmission of MDROs during their hospital stay.