As the old population continues to grow, their proportion of the total Korean population is also rapidly increasing [1]. Comorbidities and mortality due to chronic diseases increase with age. Advances in diagnostic and treatment technology extend life expectancy. They also prolong the lives of even terminally ill patients with merely a chance of resuscitation. This life-sustaining treatment (LST) may increase patient’s suffering and economic burden but not relieve their debilitating symptoms. The Korean National Agency for Management of Life-Sustaining Treatment defines LST as medical treatment by cardiopulmonary resuscitation, hemodialysis, administration of anti-cancer drugs, mechanical ventilation, and other medical treatments prescribed by Presidential Decree to a patient at the end stage of life that merely extend the duration of the end stage of life without curative effects [2]. Useless LST was frequently and automatically performed against the patient’s will or at least without knowing the patient’s will in hospitals under some medical circumstances. This phenomenon may be due to the development of medical technology that can extend life, the Korean payment system as a fee-for-service, or lack of legal protection for medical staff when they do not perform useless LST for patients without the possibility of a cure [3]. The patient’s right to self-determination and to pursue happiness may be at odds with the medical team’s obligation to maintain life. Along with medical, ethical and legal problems that may arise in this process, it is also concerning that the quality of life of terminally ill patients will deteriorate due to prolonged LST. For these complex reasons, the LST Decision Act was legislated in 2016 and has been implemented in Korea since 2018 [4].

Therefore, it is necessary to know which factors affect advance care planning (ACP) such as LST preference for old community-dwelling Koreans. The purpose of this study was to investigate which factors are associated with LST disagreement in the old population based on the 2017 National Survey of Older Persons (NSOP).

1. Data source

The 2017 NSOP is a cross-sectional and national representative survey conducted by the Ministry of Health and Welfare and the Korea Institute for Health and Social Affairs (KiHASA) [5]. It is taken every three years to understand the living conditions, needs, and desires of old community-dwelling Koreans and improve their quality of life [5]. The NSOP provides basic information to make policies for Korea’s old population. To maintain a representativeness of the Korean old population, the survey used two-stage stratification extraction for applying to sampling weights. It was conducted on ten people in urban areas and 20 people in rural areas for each survey, and more than 400 participants were required to complete the survey to calculate statistics by city and province. As a result, surveys were conducted in seven metropolitan cities, nine small to medium-sized cities, and nine rural areas.

2. Study population

The survey was conducted between June 12, 2017 and August 28, 2018 by 60 professional surveyors who were trained by research staff in advance. The NSOP was conducted through a direct interview survey of households with residents aged 65 or older in the pre-sampled area. The preference for LST questions were asked among 10,299 people aged 65 or older. Among them, 9,699 were finally included after excluding individuals who responded by proxy (n=216), individuals who were diagnosed with cancer (n=360), and individuals who had missing data (n=24) (Figure 1). This study was approved by the Institutional Review Board of Chungbuk National University Hospital (IRB No. CBNUH2021-06-025). This research followed the 1964 Helsinki Declaration.

Figure 1. Flowchart of inclusion and exclusion.

3. Definition of major variables

The questionnaire about LST preference asked “What do you think about medical practice (life-sustaining treatment) to save you even though you are unconsciousness or difficult to live?” The answers were organized on a five-point Likert scale from “strongly agree”, “agree”, “neutral”, “disagree” to “strongly disagree”. The answer of “strongly disagree” was designated as ‘Disagreement’ for LST and the others were assigned to ‘others.’

Demographic (age and sex), socio-economic (household structure, education, personal income, and private insurance), behavioral (smoking, alcohol, and exercise), and medical (cognitive function, depression, and hospitalization) characteristics were used in the analysis as variables that affect LST preference. Variables except for age and annual personal income were dichotomized as categorical variables. Cognitive function was classified into three cognitive statuses according to mini-mental state examination (MMSE) score: normal, ≥24; mild, 20-23; and dementia, ≤19. Household structure was classified into living alone, living with spouse, and living with other people besides spouse. Education status was stratified into three groups: short, <6 years; intermediate, 6-<12 years; and long, ≥12 years. Annual personal income was defined as the data recorded for total annual personal income as a continuous variable. Private health insurance subscribers were defined as individuals with private health insurance in addition to the obligatory Korean national medical insurance. Almost all individuals with private health insurance held both national and private medical insurance. Depression was defined as those who answered that they had ever been diagnosed by a doctor. Hospi-talization was defined as individuals who had been hospitalized for illness or injury in the past year. Current smokers and drinkers were defined as individuals who answered “I currently smoke cigarettes” and “I drink alcohol once or more a week”, respectively. Regular exercisers were defined as individuals who were engaged in at least 150 minutes per week.

4. Statistical analysis

Continuous and categorical variables were presented as means±standard errors (SEs) and weight number (percentage), respectively. All sampling and weight variables were stratified. P-values were calculated by multiple regression analyses after applying weighting. Frequency analysis and technical statistics analysis were conducted to identify the general characteristics and the degree of variables of the respondents. Chi squared tests were conducted to identify the preference for LST, general characteristics, and association with variables. Multiple logistic regression analyses were performed to calculate adjusted odds ratios (ORs) and 95% confidence intervals (CIs) after adjusting for factors affecting the preference and correlation of variables (age, cognitive function, household structure, education period, economic status, subscription to private insurance, depression, hospitalization, smoking status, drinking status, and exercise status). Statistical analyses were conducted using the statistical package SAS enterprise version 7.1 (SAS Inc., Cary, NC, USA). All P-values were based on two-sided tests and P-values below 0.05 were regarded as statistically significant.

Table 1 describes participant characteristics by sex. 3,856 males and 5,843 females were included. The mean age of males and females was 73.5 and 74.3 years, respectively. 42.9% of males and 42.5% of females strongly disagreed with LST. Mild cognitive impairment (MMSE 20-23) and dementia (MMSE ≤19) were 14.1 and 4.0%, respectively, in males and 22.4 and 13.2%, respectively, in females. 12.2% of males and 36.6% of females lived alone, while 64.2% of males and 35.6% of females lived with spouse. Old males were more educated and had higher personal income than old females. Mean annual personal income was 17.118 million KRW for males and 8.002 million KRW for females. Percentages of depression and hospitalization within the past year were higher in females than in males. Current smokers, drinkers and regular exercisers were higher in males than in females.

Table 1 . Participant characteristics by sex.

	Males	Females	P-values
Unweighted Number	3,856	5,843
Age, years	73.5±0.1	74.3±0.1	<0.001
LST Disagreement, weighted N (%)	1,678.5 (42.9)	2,395.6 (42.5)	0.724
MMSE score, weighted N (%)			<0.001
Normal (≥24)	3,204.6 (81.9)	3,626.3 (64.4)
Mild (20-23)	550.2 (14.1)	1,263.7 (22.4)
Dementia (≤19)	158.3 (4.0)	742.6 (13.2)
Household structure, weighted N (%)			<0.001
Alone	477.2 (12.2)	2,059.1 (36.6)
With spouse	2,513.0 (64.2)	2,007.5 (35.6)
With others besides spouse	922.9 (23.6)	1,566.0 (27.8)
Education, weighted N (%)			<0.001
Short (<6 years)	396.2 (10.1)	1,959.1 (34.8)
Intermediate (6-<12 years)	2,045.5 (52.3)	2,840.0 (50.4)
Long (≥12 years)	1,471.3 (37.6)	833.5 (14.8)
Personal income, 103 KRW/year	1,711.8±33.8	800.2±11.7	<0.001
With private insurance, weighted N (%)	1,185.8 (30.3)	1,669.9 (29.7)	0.490
Depression, weighted N (%)	76.8 (2.0)	220.3 (3.9)	<0.001
Hospitalization, weighted N (%)	560.0 (14.3)	981.0 (17.4)	<0.001
Current smokers, weighted N (%)	803.7 (20.5)	167.4 (3.0)	<0.001
Current drinkers, weighted N (%)	1,841.9 (47.1)	698.4 (12.4)	<0.001
Regular exercise, weighted N (%)	2,174.0 (55.6)	2,400.0 (42.6)	<0.001

All variables were presented as mean or percentage±standard errors..

LST, life-sustaining treatment; MMSE, mini-mental state examination..

P-values were calculated from chi-square tests for categorical variables or independent t-tests for continuous variables..

Table 2 demonstrates participant characteristics according to LST preference by sex. There was no age difference in males (73.3 vs 73.7, P-value=0.070), while females who strongly disagreed with LST were younger than those in the “other” category (73.9 vs 74.6 years, P-value <0.001). Females in the LST disagreement group had better cognitive functions, were more educated, had more annual personal income, and were more likely to be private insurance holders (all P-values <0.05). However, there were no factors that were significantly different between LST disagreement and others in males.

Table 2 . Participant characteristics according life-sustaining treatment preference by sex.

	Males				Females
	Disagreement	Others	P-values		Disagreement	Others	P-values
Unweighted Number	1,600	2,256			2,399	3,444
Age, years	73.3±0.2	73.7±0.1	0.070		73.9±0.1	74.6±0.1	<0.001
MMSE score, weighted N (%)			0.232				<0.001
Normal (≥24)	1,393.2 (83.0)	1811.4 (81.1)			1,655.1 (69.1)	1,971.2 (60.9)
Mild (20-23)	225.4 (13.4)	324.8 (14.5)			472.1 (19.7)	791.6 (24.4)
Dementia (≤19)	59.9 (3.6)	98.4 (4.4)			268.4 (11.2)	474.2 (14.7)
Household structure, weighted N (%)			0.132				0.511
Alone	219.3 (13.1)	258.0 (11.5)			893.2 (37.3)	1165.9 (36.0)
With spouse	1,090.6 (65.0)	1422.4 (63.7)			853.2 (35.6)	1154.3 (35.7)
With others besides spouse	368.7 (22.0)	554.2 (24.8)			649.2 (27.1)	916.8 (28.3)
Education, weighted N (%)			0.607				<0.001
Short (<6 years)	178.0 (10.6)	218.3 (9.8)			791.9 (33.1)	1,167.3 (36.1)
Intermediate (6-<12 years)	865.3 (51.6)	1180.2 (52.8)			1,203.0 (50.2)	1,636.9 (50.6)
Long (≥12 years)	635.3 (37.9)	836.1 (37.4)			400.72 (16.7)	432.8 (13.4)
Personal income, 103 KRW/year	1,719.1±49.7	1,706.3±33.3	0.831		828.8±16.8	779.1±13.0	0.019
With private insurance, weighted N (%)	534.8 (31.9)	651.0 (29.1)	0.066		782.6 (32.7)	887.3 (27.4)	<0.001
Depression, weighted N (%)	24.3 (1.5)	52.5 (2.4)	0.044		92.93 (3.9)	127.3 (3.9)	0.916
Hospitalization, weighted N (%)	239.0 (14.2)	321.0 (14.4)	0.912		428.4 (17.9)	552.6 (17.1)	0.426
Current smokers, weighted N (%)	358.1 (21.3)	445.7 (19.9)	0.288		62.6 (2.6)	104.9 (3.2)	0.170
Current drinkers, weighted N (%)	805.1 (48.0)	1,036.8 (46.4)	0.332		312.6 (13.1)	385.8 (11.9)	0.204
Regular exercisers, weighted N (%)	928.3 (55.3)	1,245.7 (55.8)	0.784		985.9 (41.2)	1,414.0 (43.7)	0.058

All variables were presented as mean or percentage±standard errors..

MMSE, mini-mental state examination..

P-values were calculated from chi-square tests for categorical variables or independent t-tests for continuous variables..

Table 3 shows a multiple logistic regression analysis to examine the factors that affect the preference of LST disagreement by sex after adjusting for possible confounders. Age was not significantly associated with LST disagreement in either sex (ORs [CIs] 0.994 [0.983-1.006] in males and 0.992 [0.982-1.002] in females). Compared with females with normal cognition (MMSE ≥24), ORs (CIs) for LST disagreement of females with mild cognitive impairment (MMSE 20-23) and with dementia (MMSE≤19) were 0.729 (0.632-0.841) and 0.707 (0.586-0.853), respectively. Compared with females who were more educated (≥12 years), ORs (95% CIs) of females educated for 6 to 11 years were 0.837 (0.713-0.981) while those of females educated for less than six years were not significant. Compared with private insurance holders, females without private insurance were more likely to choose LST disagreement (OR [95% CIs] 1.151 [1.006-1.316]). Females who regularly exercised were less likely to strongly disagree with LST than females who did not (0.817 [0.731-0.913]). Males who lived with others besides spouse were less likely to select LST disagreement compared with males who lived alone (0.757 [0.603–0.949]).

Table 3 . Multiple logistic regression models for life-sustaining treatment disagreement by sex.

Odd ratios (95% confidence intervals)	Males	Females
Age, every 1 year	0.994 (0.983-1.006)	0.992 (0.982-1.002)
MMSE score
Mild impairment (20-23) vs normal (≥24)	0.914 (0.755-1.107)	0.729 (0.632-0.841)
Dementia (≤19) vs normal (≥24)	0.778 (0.549-1.102)	0.707 (0.586-0.853)
Household structure
With spouse vs Alone	0.888 (0.727-1.085)	0.885 (0.771-1.014)
With others besides spouse vs Alone	0.757 (0.603-0.949)	0.900 (0.785-1.033)
Education
<6 years vs ≥12 years	1.172 (0.921-1.491)	0.899 (0.745-1.084)
6-11 years vs ≥12 years	0.980 (0.851-1.128)	0.837 (0.713-0.981)
Personal income, every 105 KRW	1.000 (1.000-1.000)	1.000 (1.000-1.000)
Private insurance (Yes vs No)	1.112 (0.955-1.295)	1.151 (1.006-1.316)
Depression (Yes vs No)	0.613 (0.376-1.001)	0.989 (0.751-1.304)
Hospitalization (Yes vs No)	1.009 (0.839-1.212)	1.079 (0.937-1.242)
Currently smokers (Yes vs No)	1.053 (0.896-1.237)	0.788 (0.572-1.085)
Currently drinkers (Yes vs No)	1.022 (0.896-1.166)	1.063 (0.903-1.251)
Regular exercisers (Yes vs No)	0.963 (0.845-1.098)	0.817 (0.731-0.913)

MMSE, mini-mental state examination..

This study demonstrated that cognitive impairment, education, private insurance subscription, and regular exercise were significantly associated with LST disagreement in females, while only household structure was associated with LST disagreement in males.

Death is inevitable and gives individuals the greatest fear. However, this fear can be reduced to prepare for and understand death properly. The process of preparing for death, such as ACP, differs between societies and cultures [6]. A multidisciplinary Delphi panel defined ACP as “a process that supports adults at any age or stage of health in understanding and sharing their personal values, life goals, and preferences regarding future medical care” [7] The purpose of ACP is to make decisions or choose preferences about the healthcare that individuals would like to receive if they happen to become unable to communicate with their family and healthcare providers. ACP includes advance directives, LST planning and use of hospice-palliative care. Thus, ACP helps care-givers and medical staff decide whether or not to provide LST for unconscious patients in incurable terminal status. Persons with ACP were more likely to die at home and less likely to receive LST such as cardiopulmonary resuscitation, intubation, and mechanical ventilation at end-of-life [8]. For these reasons, ACP appears to be cost-effective and improve quality of life of patients and their family to avoid useless and expensive LST [9,10]. Previous studies reported factors that affect preference of ACP and LST for terminally ill patients and the community-dwelling old population [11,12]. For both terminally ill patients and community-dwellers, the most important factor in determining the preference of LST is to understand whether the disease is curable [11,12]. An Australian study reported that patients with higher frailty, those living in an institution, and those with recent hospitalization were more likely to prefer avoiding LST [13].

However, although several studies have reported factors regarding preference of ACP or LST planning, there is little research that examines the association between LST preference and a wide range of variables including demographic (age and sex), socio-economic (household structure, education, personal income, and private insurance), behavioral (smoking, alcohol intake, and exercise), and medical (cognitive function, depression, and hospitalization within one year) characteristics. We hypothesized that socio-economic and medical conditions can affect LST preference in the old population. This study demonstrated that cognitive decline including mild impairment and dementia, intermediate education (6-11 years), and regular exercise were negatively associated with LST disagreement in females, while female private insurance holders were more likely to disagree with LST. Unlike in females, only household structure was associated with LST disagreement in males. LST preference and ACP documentation appear to be influenced by individual preference, socioeconomic status, health status, the medical system, culture, and the legal system in a complex way. For instance, Khosla et al. reported that education and socioeconomic status were rarely associated with ACP engagement in US adults [14], whereas Nouri et al. presented that neighborhood socioeconomic status was negatively associated with ACP documentation in San Francisco, California [15]. In this study, personal income was not related to LST preference in either sex. However, female private insurance holders were more likely to strongly disagree with LST. This result is opposite to our expectation. We expected that the private insurance holders might prefer more aggressive and expensive treatment even if they were incurable. This result is thought to be due to the characteristics of people who subscribe to private insurance in addition to the Korean national health insurance, not the coverage issue of private insurance. Private insurance holders are more likely to want to avoid compromising their quality of life at the end stage of life. Cognitive impairment in the old population is challenging medical decision-making about patients’ health care [16]. Thus, knowing LST preference or documentation of ACP plays an important role in deciding whether LST will be done as cognitive function declines. Despite the importance, there is a lack of research on the association between cognitive function and LST preference in community-dwelling old people. This study demonstrated that old females with cognitive impairment were less likely to strongly disagree with LST. As cognitive decline progresses, autonomous decision-making may become more difficult. Thus, if active discussion with old people with normal cognitive function about their own ACP and LST plan is encouraged, futile LST may be reduced when they have a terminal illness and cognitive decline. Low educational status was negatively associated with LST disagreement. People with low education may be less likely to understand the meaning of ACP and LST. Thus, education regarding hospice-palliative care and LST may decrease useless LST for patients with incurable advanced diseases [17]. Regular exercise was inversely related to LST disagreement. Persons who exercise regularly are confident in their health and do not seem to give up LST even under the assumption of a terminal disease.

There are several limitations to this study. This study is cross-sectional research. It is not possible to confirm the causality between LST preference and various factors. In addition, we do not know whether individuals who strongly disagree with LST actually forgive futile LST at the end-of-life when they have an incurable disease. Furthermore, care-givers tend to not inform patients of their terminal condition and make legal decisions about the medical treatment instead of patients them-selves when the patients have advanced diseases but are even conscious in Korea. These unique features of the Korean culture make it easy to violate patient’s autonomy. The survey regarding LST plan was performed in community-dwelling old people, not in hospitalized patients with advanced diseases. If we used a legal form such as ACP or advance directive for hos-pitalized patients, we would be able to investigate more precise associations. We defined LST disagreement as individuals who responded that they strongly disagreed with LST. From the raw data analysis, the percentage of LST disagreement, including disagreement and strong disagreement, was 91.8%. Because we thought the percentage was overestimated, LST disagreement was defined as only strong disagreement with LST.

However, this research has several advantages. The study population is representative of old Korean people dwelling in community. To represent the general Korean population aged 65 or older, we applied sampling weights to all analyses. In addition, comprehensive variables including demographic, socio-economic, behavioral, and medical factors were adjusted to examine their association with LST preference. Thus, the findings from this study can be generalized to represent the old Korean population and used for policy development.

In conclusion, cognitive decline, low educational status, and regular exercise were negatively associated with LST disagreement, while private insurance subscription was positively related to LST disagreement in females. Only household structure was associated with LST disagreement in males.

This work has received a research grant from the National Research Foundation of Korea (Grant No. 2021R1G1A1006485)

No potential conflict of interest relevant to this article was reported.

HTK designed this research, got the data and wrote the manuscript: SJS and HCL conducted data management and statistical analyses. All authors read and approved the final manuscript.