INTRODUCTION
Ensuring patients’ right to self-decision is an important issue in end-of-life care [1]. The Act on Decisions on Life-Sustaining Treatment, enacted in 2016, provides a legal basis to guarantee patients’ best interests and self-decision [2]. According to this law, a decision to withhold or withdraw life-sustaining treatment (LST) refers to withholding or discontinuing medical interventions that only prolong life without therapeutic benefit for terminally ill patients. Patients may make such decisions through either an Advance Directive (AD) or a LST Plan at the terminal stage of disease trajectory (i.e., self-decision). However, if the patient is unable to make decisions and has no documented intention, the patient’s family may make proxy decisions in the end-of-life stage (family decision). Based on self-decision or family decisions, two physicians determine the end-of-life stage and may proceed with the decision to withhold or withdraw LST accordingly [2].
Despite social and political efforts, the rates of AD or a LST Plan are still low, and in particular self-decision rate is notably low. In 2023, only 20.1% of all deaths involved the implementation of such decisions, and the rate of self-decision among them was also low at 45.0%, which shows that more than half of the cases followed the decisions of family members [3]. In 2023, the proportion of older adults aged 65 and above who completed AD was only 16.0% [3,4]. This finding suggests that most older adults are making self-decisions through LST plan in hospitals [3].
As Korea enters a super-aged society, the low rate of self-decision among older adults is a social issue that cannot be overlooked. In Korean culture, illness is often regarded as a family matter. For older patients, family members frequently make medical decisions instead of the patients themselves [5]. These cultural characteristics can restrict self-decision, increase the psychological and economic burdens on families, and lead to unnecessary increases in LST, thus increasing the national healthcare burden [6,7]. Therefore, research aimed at strengthening the self-decision in older adults, while considering their specific characteristics, is needed in the field of gerontological nursing. Because the LST plan represents the last opportunity to secure self-decision in older adults who have not completed an AD, research on this topic is particularly necessary [1,2].
Previous studies primarily focus on the current status of LST [7-9] and the contributing factors of AD [10]. Most of these studies are conducted in community settings with participants from various age groups [7-9]. Hospital-based research is limited to specific clinical environments, such as emergency departments, intensive care units, and trauma centers [8]. Because practices related to LST differ across countries, research tailored to each nation’s cultural context is necessary [11]. In Korea, where 75.4% of all deaths occur in healthcare institutions [4], studies focusing on hospitalized older adults are needed to identify factors that influence self-decision in LST plan.
A large volume of Korean studies have investigated self-decision; however, those studies did not make a clear distinction between AD and LST plan [5,6,12,13], or focused on the AD [14]. Furthermore, these studies predominantly analyzed the data shortly after the initial Act on LST [5,6,13] and targeted adult populations, particularly cancer patients [6,12]. As a result, research examining factors that affect self-decision through LST plan among older adults remains insufficient. Previous domestic studies on self-decision [5,6,12,13,15] have mainly explored the effect of demographic characteristics on patient self-decision. These studies report that male subjects[5], individuals younger than 65 years [5,15], and cancer patients [6] are more likely to do self-decision, whereas individuals with disabilities, those residing outside metropolitan areas [5,6], those with lower income levels [5,6], and intensive care unit patients [13] are less likely to do so [5,6]. In contrast, limited studies included clinical factors reflecting the patient’s conditions during the early period of hospitalization.
Nursing assessments during the early period of hospitalization (e.g., level of consciousness, fall risk, and pressure ulcers) [16], clinical laboratory indicators (e.g., elevated C-reactive protein (CRP) levels, hypoalbuminemia, and increased blood urea nitrogen) were reported to be significant predictors of in-hospital mortality and survival outcomes [17]. However, research examining nursing assessments and clinical laboratory results in relation to decisions regarding LST is limited. Only few studies have compared clinical laboratory results between patient self-decision and family-decision groups among adult patients [18] and have analyzed symptoms (e.g., pain and dyspnea), and analgesic use among older long-term care residents [19]. Moreover, research investigating the association between early hospitalization clinical status and self-decision among older adult patients remains insufficient.
With the revision of the Act on Decisions on Life-Sustaining Treatment in 2019, the number of LST items increased from four to seven [2], which may broaden the scope of self-decision and reduce unnecessary medical interventions. However, most previous studies were conducted before this legislative amendment [5,6,13,15,20] and therefore, do not adequately reflect changes in self-decision patterns after the revision.
Therefore, we aim to analyze factors that affect self-decision in LST decision in hospitalized older adult patients during their final hospitalization before death, using data collected after the legislative amendment. We expect our findings to provide evidence for identifying appropriate intervention timing and developing effective strategies to support self-decision among older adults.
The specific objectives of this study are as follows:
1) To identify differences in general characteristics, clinical characteristics, laboratory results, and nursing assessment findings between the self-decision group and the family-decision group at admission.
2) To identify differences between the self-decision group and the family-decision group in clinical characteristics and in withholding or withdrawing LST items at the time of completing the Form for Implementation of LST decisions (LST implementation form).
3) To analyze differences in healthcare service utilization between the self-decision and the family-decision groups after completion of the LST implementation form.
4) To identify independent factors contributing to the self-decision based on characteristics at admission.
METHODS
Ethics statement: This study was approved by the Institutional Review Board (IRB) of the Catholic University of Korea (IRB No. UC23RISI0080). Informed consent was waived for the participants.
1. Study Design
This study is a retrospective case-control study comparing the self-decision group and the family-decision group, aimed at identifying factors that influence self-decision among hospitalized older adult patients. We have reported the study in accordance with the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guideline (https://www.strobe-statement.org).
2. Study Participants
The participants in this study were older patients aged 65 years or older who died at a Catholic University Hospital in Gyeonggi Province between January 1, 2020, and December 31, 2022, and who met the following inclusion criteria. The inclusion criterion for the self-decision group was those who completed a LST plan after admission. The inclusion criterion for the family-decision group were completing either a “confirmation of patient’s intention for withholding or withdrawing LST” (provided by statements from at least two family members) or a “confirmation of intention by guardians and all family members” (consensus among all patient family members) after admission. Exclusion criteria were patients who died within one day of admission, those for whom the LST implementation form had been completed during a previous admission, or those whose the LST implementation form or medical records were incomplete (Figure 1).
During the study period, 1,977 older adults aged 65 years or older died at the hospital. The final sample was 624 participants: 249 in the self-decision group and 375 in the family-decision group. In the family-decision group, 166 cases were identified by statements from at least two family members, and 209 were determined by consensus among all family members (Figure 1). The sample size exceeded the minimum required sample size of 215 individuals, which we estimated using G-Power 3.1.9.7, ensuring an adequate sample size for this study. We estimated the required sample size for logistic regression analysis based on a previous study [5], using an odds ratio (OR) of 2.99, a significance level of .05, a power of .95, and a two-tailed test.
3. Study Variables
1) Characteristics at admission: general characteristics, clinical characteristics, laboratory results, and nursing assessment findings
General characteristics include age, sex, religious affiliation (yes or no), type of insurance, medical department, route of admission (outpatient or emergency room), and admission ward (general ward or intensive care unit). We categorized age groups as 65~74 years (young-old), 75~84 years (old-old), and 85 years or older (oldest-old). Insurance types are classified as National Health Insurance and Medical Aid. Medical departments are divided into medical, surgical, and emergency/trauma specialties.
Clinical characteristics included underlying diseases, comorbidities, and laboratory results. We classified underlying diseases using the main diagnostic code at death, which included malignant neoplasms, circulatory system diseases, respiratory system diseases, digestive system diseases, genitourinary system diseases, and other diseases. We also recorded whether comorbidities were present or absent.
Laboratory test results were based on the first tests after admission. Using the reference ranges provided in the diagnostic reports and previous research [17], we categorized results as normal or abnormal. We also used the modified Glasgow Prognostic Score (mGPS), which was developed to predict survival based on inflammation and nutritional status in cancer patients. This score incorporates recent evidence showing that hypoalbuminemia alone does not always indicate a poor prognosis. We calculated the mGPS as follows: 0 points for CRP ≤1 mg/dL and albumin ≥3.5 g/dL, 1 point for CRP >1 mg/dL, or 2 points for CRP >1 mg/dL and albumin <3.5 g/dL. In this study, the investigator directly calculated the mGPS using clinical laboratory values at admission. Higher scores indicate a worse prognosis [21].
The nursing assessments included level of consciousness, pain evaluation, pressure ulcer risk, fall risk, and nutritional status. Level of consciousness was assessed based on documentation in the nursing information sheet at admission and was categorized as alert, drowsy, stuporous, or semicomatose. Pain was evaluated using the Numeric Rating Scale for patients who were able to communicate and understand numbers. For patients unable to communicate, we used the Face, Legs, Activity, Cry, and Consolability scale. Both tools have a scoring range of 0~10, categorized as no pain (0), mild pain (1~3), and moderate to severe pain (≥4). In this study, we classified pain levels based on the initial nursing assessment conducted within 24 hours of admission. Pressure ulcer risk was assessed using the Braden Scale during the first nursing evaluation within 24 hours of admission. The Braden Scale includes six subcategories: sensory perception, moisture, activity, mobility, nutrition, and friction/shear. The total score ranges from 6 to 23. Based on the score, patients were categorized as no risk (19~23), mild risk (15~18), moderate risk (13~14), or high risk (6~12) [22]. Fall risk was assessed using the Johns Hopkins Fall Risk Assessment Tool during the first evaluation within 24 hours of admission. This tool evaluates seven factors: history of falls, elimination, age, medications, use of patient care equipment, mobility, and cognition. The total score ranges from 0 to 35, with a higher score indicating a greater risk of falls. Patients were classified as low risk (0~5), moderate risk (6~13), or high risk (14~35) [23]. We evaluated nutritional status using a computerized item developed by the hospital, based on the Nutritional Risk Screening 2002 (NRS-2002) tool. This assessment integrates initial evaluations documented by nurses and physicians within 48 hours of admission and includes eight subcategories: weight loss in the past three months, body mass index, food intake status, gastrointestinal disturbances, bowel dysfunction, activity level, dietary prescriptions, and age. The total score ranges from 0 to 24, with patients categorized as well-nourished (0~2) or at risk of malnutrition (≥3) [24].
2) Characteristics and withholding or withdrawing LST items at the time of completing the LST implementation form
Clinical characteristics collected at the time of LST implementation form included the number of days from hospital admission, the location where form occurred (general ward, intensive care unit, emergency department, or hospice ward), the presence of a DNR (do-not-resuscitate) consent form, and a history of surgery. We also collected the data on the withholding or withdrawing LST items: cardiopulmonary resuscitation, mechanical ventilation, hemodialysis, chemotherapy, extracorporeal life support, blood transfusion, and vasopressor use.
3) Healthcare service utilization after completion of the LST implementation form
Variables related to healthcare service utilization after completion of the LST implementation form included total length of hospital stay (from admission to death), time from document completion to death, death occurrence in the intensive care unit, use of hospice and palliative care services, and use of invasive medical devices. The invasive devices included urinary catheter, feeding tubes (nasogastric and gastrostomy tubes), percutaneous transhepatic biliary drainage catheter, central venous catheter, chemoport, peripherally inserted central catheter, arterial catheter, hemodialysis catheter, mechanical ventilator, and tracheostomy tube [25]. All devices present at the time of death were included, regardless of the timing of insertion.
4. Ethical Considerations and Data Collection Procedure
This study received approval from the Institutional Review Board of Catholic University Hospital in Gyeonggi-do (IRB No. UC23RISI0080), and the requirement for informed consent was waived. We obtained approval from the medical institution to transfer medical records for research purposes. All personal identifiers were removed from the data provided by the medical records department, and both the data set and case report forms were securely stored on a password-protected computer. Two researchers collected additional data necessary for the study from electronic medical records using structured case report forms between October 1, 2023, and January 31, 2024.
5. Data Analysis
We analyzed the data using SPSS Windows version 22 (IBM Corp.). We examined differences in general characteristics, clinical characteristics, laboratory findings, and nursing assessments between the self-decision and the family-decision groups using frequencies, percentages, means, standard deviations, chi-square tests, and independent-sample t-tests. We analyzed differences in clinical characteristics and withholding or withdrawing LST items at the time of LST implementation form between the two groups using chi-square tests, Fisher’s exact tests, and independent-sample t-tests. We examined differences in healthcare service utilization after documentation using chi-square tests and independent-sample t-tests. We analyzed factors that affect self-decision during the process of LST implementation using binary logistic regression.
RESULTS
1. Differences in Characteristics between Self-Decision and Family-Decision Groups at Admission
In the present study sample, 39.9% (249 patients) were in the self-decision group, and 60.1% (375 patients) were in the family-decision group. We found significant differences in general characteristics at admission between the two groups for age (t=-7.17, p<.001), religion (χ2=4.64, p=.031), medical department (χ2=45.18, p<.001), admission route (χ2=89.92, p<.001), and admission ward (χ2=117.71, p<.001). We did not find differences in sex or insurance types (Table 1).
Regarding clinical characteristics at admission, we found a significant difference only in underlying diseases (χ2=145.29, p<.001), while comorbidities did not differ between groups (Table 1). Laboratory test results at admission showed significant differences between the groups in hemoglobin (χ2=5.07, p=.024), hematocrit (χ2=12.53, p<.001), creatinine (χ2=26.64, p<.001), and mGPS (χ2=8.37, p=.015). We observed no differences in white blood cell count, lymphocyte count, blood urea nitrogen, total protein, albumin, serum glutamic-oxaloacetic transaminase, serum glutamate pyruvate transaminase, sodium, potassium, or CRP (Table 1).
In the nursing assessment results at admission, we found differences between groups in all variables. The self-decision group included a higher proportion of patients with clear consciousness (92.4% vs. 58.4%) and higher pain scores (2.35±1.98 vs. 1.37±0.07). Additionally, the self-decision group had more patients at high risk of falls (87.2% vs. 80.8%) and with poor nutritional status (47.8% vs. 37.3%). In contrast, the family-decision group included a higher proportion of patients classified as no risk for pressure ulcers (41.8% vs. 15.7%).
2. Differences in Clinical Characteristics and Withholding or Withdrawing LST Items at the Time of Completing the LST Implementation Form
At the time of completing the LST implementation form, the two groups differed in the location of form completion (χ2=229.50, p<.001), DNR consents (χ2=62.66, p<.001), and surgery history (χ2=29.81, p<.001). The average period from admission to completion of the LST implementation form was 7.74 days in the self-decision group and 9.32 days in the family-decision group, with no significant difference between the groups (t=-1.92, p=.056).
The self-decision group was more likely to withhold or withdraw all seven types of medical interventions compared to the family-decision group (63.5% vs. 22.1%; χ2=107.78, p<.001). In the detailed analysis, the self-decision group showed higher rates of withholding or withdrawing interventions, such as mechanical ventilation, hemodialysis, chemotherapy, extracorporeal life support, blood transfusion, and vasopressors, except for cardiopulmonary resuscitation (Table 2).
3. Differences in Healthcare Service Used after Completion of the LST Implementation Form
Following completion of the LST implementation form, healthcare services used between the two groups significantly differed in the use of the intensive care unit (ICU) before death (χ2=125.24, p<.001) and hospice and palliative care services (χ2=189.22, p<.001). The self-decision group used hospice and palliative care services significantly more than the family-decision group (53.8% vs. 5.3%), while the family-decision group had a higher rate of ICU use (46.1% vs. 4.4%). We found no significant differences between the groups in total hospitalization duration or in the period from form completion to death.
Additionally, the self-decision group maintained fewer invasive medical devices-including urinary catheters, feeding tubes, central venous catheters, arterial lines, hemodialysis catheters, mechanical ventilators, and the total number of insertions (p<.001), as well as tracheostomy tubes (p=.033) compared to the family-decision group. Conversely, the self-decision group was more likely to maintain percutaneous transhepatic biliary drainage tubes (p=.009), chemoports (p<.001), and peripherally inserted central catheters (PICC lines) (p=.003) (Table 3).
4. Factors Influencing the Self-Decision for LST Implementation at Admission
To identify factors influencing the self-decision, we selected 14 variables that showed significant differences in bivariate analysis between the two groups at admission. These variables included age, religion, medical department, route of admission, admission ward, underlying diseases, hemoglobin, creatinine, mGPS, level of consciousness, pain assessment, pressure ulcer risk, fall risk, and nutritional status. Hematocrit was excluded from the model because it is highly correlated with hemoglobin as an indicator of blood oxygen-carrying capacity, which ensured model stability.
The logistic regression model was statistically significant (χ2=270.77, p<.001; Hosmer–Lemeshow test, p=.083), with a Nagelkerke R² value of .487. The analysis showed that age, medical department, admission ward, underlying disease, level of consciousness at admission, and pain assessment at admission were significant factors affecting self-decision. Compared to older adults aged 65~74, the likelihood of self-decision was lower in the 75~84 age group (OR=0.54, 95% confidence interval [CI]=0.34~0.85) and in those aged 85 and above (OR=0.37, 95% CI=0.19~0.73). The likelihood of self-decision increased for patients admitted to a medical department (OR=12.60, 95% CI=2.83~56.13), those hospitalized in general wards (OR=2.83, 95% CI=1.33~6.02), patients with cancer (OR=2.01, 95% CI=1.16~3.49), those with an alert level of consciousness at admission (OR=2.27, 95% CI=1.15~4.51), and those experiencing pain at admission (OR=2.64, 95% CI=1.68~4.13) (Table 4).
DISCUSSION
This study investigated differences between self-decision and family-decision groups and analyzed factors influencing self-decision among 624 older adult inpatients who completed the LST implementation form. Among the participants, only 39.9% made self-decisions in LST planning, while more than 60% relied on family decisions. Notably, over half of the family-decisioned cases involved full family consensus, suggesting that patients’ preferences were often not communicated or honored by their families.
These findings support the need for policy interventions to increase the LST planning. Currently, the timing of LST plan completion is recommended at the terminal stage [2]. However, because it is difficult to determine terminal status [11], guidelines for appropriate timing should be reestablished. Decision-making capacity often declines significantly after the terminal phase, which makes discussions on self-decision difficult [11,15]. Therefore, the second national hospice and LST policy plan (2024~2028) aims to extend the timeframe for LST planning, allowing patients to make well-considered decisions [1]. The results of this study highlight the necessity of institutional frameworks that enable older adults to complete LST plan before reaching the terminal stage, which can better safeguard their autonomy.
In this study, factors influencing self-decision include medical department, ward type, presence of pain, level of consciousness, underlying disease, and age. Medical department is the most significant predictor of self-decision; patients admitted to internal medicine are 12.6 times more likely to make self-decisions compared to those in surgical or emergency/trauma departments. Similar findings have been reported in studies involving adult and cancer patients [13,15], and internal medicine patients are known to be 21.8 times more likely to engage in LST discussions compared to surgical patients [26].
Internal medicine patients, who often manage chronic illnesses, typically experience a gradual decline in health status and have more opportunities for exposure to hospice or palliative care. In contrast, patients admitted for acute surgical or traumatic conditions often experience rapid deterioration, which leaves limited opportunity for LST discussions and increases the likelihood of family-driven decisions [7,13]. These findings suggest that counseling and education systems for patients with chronic diseases or cancer should be prioritized to facilitate earlier completion of LST plan [1]. Developing standardized guidelines will enable healthcare professionals to clearly explain disease trajectories, medical interventions, and the implications of LST decisions. Additionally, encouraging AD in anticipation of unexpected end-of-life situations is essential.
In this study, ward type also significantly affected self-decision; patients admitted to general wards were more likely to make self-decisions than those in ICUs. In ICUs, aggressive treatments often take precedence, which limits opportunities for patient decision-making [12]. However, studies show that introducing end-of-life protocols and integrating palliative care consultations within ICUs can increase LST discontinuation rates and transfers to general wards [25,27]. These findings emphasize the need for standardized protocols and structured communication pathways to facilitate LST discussions early during ICU admission [27].
Among initial nursing assessments, patients who were alert or experiencing pain had a higher likelihood of self-decision. Previous studies report no significant association between self-decision and initial consciousness [13] or pain experienced after LST decisions [19]. Because of the limited number of studies and variations in data collection timing, drawing definitive conclusions about these variables remains challenging. Nonetheless, our findings suggest that being alert and able to report pain may indicate readiness for self-decision. Future research should examine whether nursing assessments can guide optimal timing for initiating LST planning discussions.
In this study, patients with cancer as an underlying disease were more likely to make self-decisions compared to those with non-cancer conditions. This finding aligns with previous research that reports higher rates of self-decision among cancer patients [15,20,28]. Cancer generally allows for clearer prognostic assessments and terminal-stage identification [13,15], whereas non-cancer illnesses, such as heart failure, chronic obstructive pulmonary disease, and chronic kidney disease, often fluctuate between exacerbation and remission, which complicates terminal stage recognition. Traumatic conditions, characterized by sudden deterioration, also present challenges for timely LST planning. Therefore, establishing more specific terminal-stage criteria for non-cancer illnesses is needed to conduct earlier interventions [8,13,15].
Increasing age was associated with a lower likelihood of self-decision in this study, with reductions of 46.0% and 63.0% for older and oldest-old patients, respectively, compared to younger-old patients. This result is consistent with the findings of previous research [5,15,29]. Multiple factors, including medical, cognitive, social, and systemic issues, may make self-decision in LST planning particularly challenging for the oldest-old [8,30]. Therefore, policy efforts should focus on encouraging AD among older adults before they reach advanced age and serious illness stages. Effective strategies may include repeated explanations through verbal discussions, printed materials, and videos, in addition to integrated communication approaches that involve family members [30].
In this study, our findings on medical interventions and healthcare utilization patterns are consistent with previous studies on end-of-life care choices [6,8,12,20]. We found that self-decision is associated with withholding or withdrawing a greater number and proportion of LST items. Additionally, self-decision is linked to higher use of hospice and palliative care services and lower ICU admissions. Patients in the self-decision group also had fewer invasive devices at the time of death. These results show that self-decision leads to a more comprehensive approach to withholding or withdrawing LST [8,9].
This study has several limitations. It was a retrospective, single-center study based on medical records, which limits generalizability. Psychological, sociocultural, and healthcare provider factors that influence decision-making were not considered [30]. Multi-center prospective studies are needed in the future to comprehensively identify factors that influence self-decision across different healthcare settings. Additionally, we assumed that documented decisions were fully implemented; however, previous research [12] shows the need to distinguish between withholding and withdrawing treatment to capture patients’ intentions more accurately and to better reflect real-world practices. Despite these limitations, this is the first study to focus specifically on older adult inpatients making LST implementation decisions during their final hospitalization. We identified factors that influence self-decision in a tertiary hospital setting and offer practical and policy implications for developing strategies to promote patient autonomy among older patients.
CONCLUSION
In this study, more than 60% of older inpatients completed LST implementation through family decision. Self-decision significantly affected their end-of-life healthcare use. We found that self-decision was more likely among patients admitted to internal medicine departments, those hospitalized in general wards, patients diagnosed with cancer, alert patients, those experiencing pain, and those of younger age. To promote self-decision among older patients, healthcare systems should be established to support early LST discussions beginning at hospital admission, focusing on patient groups with higher decision-making potential, such as internal medicine patients, cancer patients, those in general wards, and conscious patients. In addition, standardized counseling and educational programs tailored to older adults are needed. Further research should examine strategies for developing policies that enable completion of LST plan before the terminal stage to strengthen the autonomy of older patients.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
