Potential Risk Factors for Lonely Death: A Nationwide Case-Control Study in Korea Authors: Hye Yeon Koo, Haibin Bai, Jae-ryun Lee, Min Jung Kang, Young-Ho Jun, Jieun Yun, Jee Hoon Sohn, Jin Yong Lee, Hyejin Lee Published: 2025 ABSTRACT BackgroundLonely death, referred to the death of an individual at home alone and found after some time has elapsed, has emerged as a significant social issue in many countries. This study aimed to distinguish high-risk groups by comparing the characteristics of individuals who died of loneliness with those of the control group.MethodsThis case-control study was conducted using data from the Korea Crime Scene Investigation and National Health Insurance Service (NHIS) databases. We identified all lonely deaths that occurred in 2021 in Korea and sex- and age-matched controls were randomly selected from the NHIS database in a 1:3 ratio. Data on demographics, health conditions, and healthcare utilisation were analysed. Associations between lonely deaths and various factors were assessed using logistic regression.ResultsThis study included 3,122 lonely death cases and 9,493 controls. Individuals who experienced lonely death were more likely to be covered by medical aid, in the lowest income group, and have high (≥ 3) Charlson comorbidity index scores than the controls (30.8% vs. 4.0%, 54.5% vs. 19.2%, 14.5% vs. 8.6%, respectively). Furthermore, they exhibited a higher prevalence of chronic diseases (including diabetes and heart failure), mental health disorders (including schizophrenia and bipolar disorder), and alcohol-related diseases, along with more frequent healthcare utilisation. Most of these factors exhibited statistical or borderline significance in adjusted logistic regression analyses.ConclusionLonely death is associated with low income, multi-morbidity, underlying diseases, and frequent healthcare utilisation. Targeted intervention strategies that focus on these potentially high-risk populations may help prevent lonely deaths. INTRODUCTION Lonely death, referred to the death of an individual at home alone and found after some time has elapsed, has emerged as a significant social issue in many countries.1 The recent coronavirus disease 2019 pandemic has raised concerns about dying alone globally.2 The number of lonely deaths is anticipated to rise with the increase in single households, and this issue can be more pressing in aging countries like Korea and Japan, where a higher population of older individuals live alone.134 Notably, 3,378 lonely deaths occurred in 2021 in Korea, accounting for approximately 1% of total deaths and reflecting a 10% and 6% mean annual rate increase in males and females, respectively, during the past 5 years.5 Most instances of lonely deaths are not considered as good deaths, as they are often the result of social isolation.16 Previous studies have reported concerns among older individuals regarding lonely death, mentioning that they need someone to care for them when dying at home.78 This is of greater significance in the context of a family-oriented society in Korea. Dying without family members is considered an undesirable death among Koreans, similar to other Asian cultures.910 The Lonely Death Prevention and Management Act of Korea states that ‘a lonely death occurs to an individual who is in a state of social isolation and is disconnected from family or other people,’ and that ‘personal and social damage caused by lonely deaths should be prevented.’11 Accordingly, the Korean government published the First Master Plan for the Prevention of Lonely Deaths in 2023 and planned to undertake various efforts to identify and manage high-risk groups for lonely death.12 The identification of at-risk individuals may help implement more focused and effective policies for monitoring and support. A previous survey conducted in 2022 indicated that lonely deaths were common among males and individuals in their 50s and 60s.5 However, evidence regarding a direct comparison between lonely deaths and other deaths is lacking. Therefore, this study aimed to identify the distinguishing characteristics between individuals who experienced lonely death and controls using nationwide data, encompassing information about all lonely deaths occurred during 2021 in Korea, to determine the high-risk groups. METHODS Study design, cases, and controls A case-control design was employed to explore the risk factors for lonely deaths. Individuals who experienced lonely death were identified using the Korea Crime Scene Investigation (KCSI) data collected by the Scientific Crime Analysis System operated by the Korea National Police Agency. The KCSI database includes information on scene investigation, collection and extraction of evidence, appraisal of evidence, and crime analysis. The keywords related to ‘lonely death’ were used to identify individuals who experienced lonely death. Subsequently, identified cases were analysed and reviewed by social welfare experts to define lonely deaths occurred during 2021. We merged KCSI and National Health Insurance Service (NHIS) data. In Korea, 97% of the population is covered under NHIS, and the remaining 3% of individuals, who have difficulty paying health insurance premiums, are covered by medical aid (MA) from the government, supporting their medical expenses through taxes. Information on both groups was collected from the NHIS database.13 NHIS database maintains comprehensive health-related information for almost the entire Korean population. We provided the information on birthdate, sex, and name to the NHIS, and the KCSI and NHIS data were merged based on the birth date, sex, name, death status, and year of death to ensure that there were no duplicates. Finally, 3,122 cases of lonely deaths were identified using NHIS data. Sex- and age-matched controls were randomly selected from the NHIS database in a 1:3 ratio. Notably, 9,493 controls were selected for this study from the total study population of 12,615. We established a 5-year retrospective cohort from 2016 to 2021 to obtain more generalisable information for this study population. Measurements We collected information on demographics, insurance types, diagnoses, treatments, prescriptions, and death status and date from the NHIS data. Demographic characteristics including sex (male and female), age group (< 20, 20–44, 45–64, 65–74, and ≥ 75), insurance type (NHIS and MA), quintile income level (health insurance contribution), geographical region (Seoul, metropolitan areas, and others), and disability were extracted from the claims data. Subsequently, the Charlson comorbidity index (CCI) score for each individual was calculated based on comorbidities in the past 5 years, and scores were categorised into three groups (0, 1–2, ≥ 3). Disease prevalence over the past 5 years for individuals who experienced lonely deaths was analysed based on the major diagnostic categories and major diseases (Supplementary Table 1). Diseases were defined according to the 8th edition of the Korean Standard Classification of Diseases and Causes of Death, which is derived from the 10th edition of the International Classification of Diseases. Healthcare utilisation was analysed by categorising outpatient visits, hospital admissions, and emergency room visits. Statistical analysis We conducted a descriptive analysis to examine demographic characteristics, underlying diseases, and healthcare utilisation patterns associated with lonely deaths. Healthcare utilisation was analysed in terms of days and presented as the mean (standard deviation [SD]). Other variables are reported as percentages. We used t-test and χ2 test for continuous and categorical data, respectively, and analysis of variance for variables with multiple groups. Logistic regression analysis was used to examine the association between lonely deaths and the related variables. The results of the logistic regression analysis are presented as unadjusted odds ratio (OR) and multivariable-adjusted odds ratio (aOR) with a 95% confidence interval (CI). Statistical significance was set at P < 0.05. Statistical analyses were conducted using SAS Enterprise Guide 7.4 software (SAS Institute Inc., Cary, NC, USA). Forest plots were constructed using the R version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria). Ethics statement This study was reviewed and approved by the Institutional Review Board (IRB) of Seoul National University Bundang Hospital (IRB No. X-2406-904-901). The requirement for informed consent was waived owing to the use of anonymised data. RESULTS This study included data of 3,122 individuals who experienced lonely death and 9,493 age- and sex-matched controls. Among the individuals who experienced lonely death, 84.0% and 16.0% were male and female, respectively (Table 1). Notably, the 45–64-year age group exhibited most of lonely deaths (57.3%), followed by the 65–74-year age group (18.5%). The proportions of individuals who were covered by MA, who were in the lowest income group, who resided in urban area (Seoul or metropolitan cities), and who exhibited a high CCI score (≥ 3) were all higher in lonely death cases compared to controls (30.8% vs. 4.0%, 54.5% vs. 19.2%, 69.6% vs. 68.1%, 14.5% vs. 8.6%, respectively; all P < 0.001). Table 1Characteristics of individuals who experienced lonely death and controlsVariablesLonely death casesControlsP valueTotal3,122 (100.0)9,493 (100.0)Sex0.929Male2,621 (84.0)7,978 (84.0)Female501 (16.0)1,515 (16.0)Age, yr1.000< 202 (0.1)6 (0.1)20–44395 (12.7)1,203 (12.7)45–641,790 (57.3)5,438 (57.3)65–74579 (18.5)1,763 (18.6)≥ 75356 (11.4)1,083 (11.4)Insurance type< 0.001NHIS2,161 (69.2)9,111 (96.0)MA961 (30.8)382 (4.0)Income level< 0.0011st (highest)178 (5.7)2,687 (28.3)2nd260 (8.3)2,021 (21.3)3rd420 (13.5)1,605 (16.9)4th543 (17.4)1,230 (13.0)5th (lowest, MA)1,702 (54.5)1,825 (19.2)Missing19 (0.6)125 (1.3)Region< 0.001Seoul1,468 (47.0)4,680 (49.3)Metropolitan cities704 (22.5)1,788 (18.8)Others950 (30.4)3,025 (31.9)Charlson comorbidity index< 0.00101,609 (51.5)5,525 (58.2)1–21,060 (34.0)3,154 (33.2)≥ 3453 (14.5)814 (8.6)Disability< 0.001Yes421 (13.5)934 (9.8)No2,701 (86.5)8,559 (90.2)Values are presented as number (%).NHIS = National Health Insurance Service, MA = medical aid. Characteristics of individuals who experienced lonely death and controls Values are presented as number (%). NHIS = National Health Insurance Service, MA = medical aid. As for underlying diseases, the prevalence of major diagnostic categories differed between the lonely death group and control (Table 2). The following disease categories were significantly more prevalent in the lonely death group than in the control: endocrine, nutritional, and metabolic diseases (67.0% vs. 61.4%), circulatory system diseases (65.3% vs. 54.8%), mental and behavioral disorders (51.4% vs. 32.0%), nervous system diseases (50.4% vs. 38.6%), factors influencing health status and contact with health services (41.0% vs. 33.0%), diseases of the blood and blood-forming organs, and certain disorders involving immune mechanisms (26.3% vs. 13.4%) (All P < 0.001). Table 2Distribution of underlying diseases among individuals who experienced lonely death and controlsDisease categoryName of diseasesLonely death casesControlsP valueMajor diagnostic categoriesDiseases of the digestive system2,776 (88.9)9,043 (95.3)< 0.001Symptoms, signs and abnormal clinical and laboratory findings2,416 (77.4)7,459 (78.6)0.038Diseases of the musculoskeletal system and connective tissue2,338 (74.9)8,172 (86.1)< 0.001Diseases of the respiratory system2,252 (72.1)8,191 (86.3)< 0.001Injury, poisoning and certain other consequences of external causes2,158 (69.1)6,868 (72.3)< 0.001Endocrine, nutritional, and metabolic diseases2,091 (67.0)5,833 (61.4)< 0.001Diseases of the circulatory system2,038 (65.3)5,206 (54.8)< 0.001Diseases of the skin and subcutaneous tissue1,997 (64.0)6,941 (73.1)< 0.001Certain infectious and parasitic diseases1,663 (53.3)5,941 (62.6)< 0.001Mental and behavioral disorders1,606 (51.4)3,042 (32.0)< 0.001Diseases of the nervous system1,575 (50.4)3,666 (38.6)< 0.001Diseases of the eye and adnexa1,556 (49.8)5,953 (62.7)< 0.001Diseases of the genitourinary system1,288 (41.3)4,236 (44.6)< 0.001Factors influencing health status and contact with health services1,279 (41.0)3,128 (33.0)< 0.001Diseases of the ear and mastoid process975 (31.2)3,400 (35.8)< 0.001Diseases of the blood and blood-forming organs and certain disorders involving the immune mechanism820 (26.3)1,272 (13.4)< 0.001Neoplasms710 (22.7)2,773 (29.2)< 0.001Congenital malformations, deformations and chromosomal abnormalities36 (1.2)157 (1.7)0.053Pregnancy, childbirth, and the puerperium9 (0.3)80 (0.8)0.002Certain conditions originating in the perinatal period1 (0.0)5 (0.1)1.000Major diseasesDyslipidemia1,669 (53.5)4,915 (51.8)0.209Hypertension1,548 (49.6)3,822 (40.3)< 0.001Diabetes1,275 (40.8)2,795 (29.4)< 0.001Hepatitis587 (18.8)1,084 (11.4)< 0.001Chronic obstructive pulmonary disease497 (15.9)1,493 (15.7)0.939Ischemic heart disease448 (14.3)1,172 (12.3)0.006Heart failure350 (11.2)517 (5.4)< 0.001Cirrhosis258 (8.3)123 (1.3)< 0.001Cancer245 (7.8)645 (6.8)0.064Stroke125 (4.0)192 (2.0)< 0.001Hip fracture60 (1.9)46 (0.5)< 0.001Dementia162 (5.2)369 (3.9)0.003Schizophrenia150 (4.8)93 (1.0)< 0.001Mood disorder1,020 (32.7)1,264 (13.3)< 0.001Depression920 (29.5)1,061 (11.2)< 0.001Bipolar disorder350 (11.2)194 (2.0)< 0.001Alcohol related diseaseMental and behavioral disorders due to use of alcohol611 (19.6)144 (1.5)< 0.001Alcohol use24 (0.8)2 (0.0)< 0.001Alcoholic liver disease689 (22.1)398 (4.2)< 0.001Alcohol related complications167 (5.3)48 (0.5)< 0.001Values are presented as number (%). Distribution of underlying diseases among individuals who experienced lonely death and controls Values are presented as number (%). Assessment of specific major health conditions revealed a higher prevalence of all diseases, such as hypertension, diabetes, heart failure, stroke, hip fracture, and cirrhosis, in the lonely death group than in the control group (Table 2). Notably, cases of lonely death were significantly associated with a higher prevalence of mental health and alcohol-related disorders than controls. Mood disorders were identified in 1,020 (32.7%) individuals who experienced lonely death and 1,264 (13.3%) controls (P < 0.001). Mental and behavioural disorders associated with alcohol use were reported in 611 (19.6%) individuals who experienced lonely death and 144 (1.5%) controls (P < 0.001). A similar trend was observed for alcoholic liver disease (22.1% vs. 4.2%, P < 0.001). As for healthcare utilisation, outpatient visits were more frequent among individuals in the lonely death group than the control group. The average days of visits were 15.4 (SD, 25.8) and 14.6 (SD, 19.6) in the lonely death group and control group, respectively, in the year 5 years prior to death (P = 0.053), and 17.4 (SD, 27.3) and 15.2 (SD, 21.4) in the year 1 year prior to death (P < 0.001). The trend remained the same for the year before death; however, in the 3 months preceding death, outpatient visits slightly decreased to 4.0 (SD, 7.1) and 3.8 (SD, 5.8) in the lonely death and control groups, respectively, reducing the difference. The length of inpatient visits was consistently longer in the lonely death group than in the control group. The average length of visits was 11.9 (SD, 49.2) days and 3.2 (SD, 25.1) days in the lonely death group and control group, respectively, in the year 5 years prior to death (P < 0.001), and 10.0 (SD, 34.1) days and 3.9 (SD, 27.6) days in the year 1 year prior to death (P < 0.001). Emergency visits were also more frequent among individuals who experienced lonely death, with the average being 1.0 (SD, 6.9) visits and 0.3 (SD, 2.6) visits in the lonely death and control groups, respectively, in the year 5 years prior to death (P < 0.001). This number was 1.8 (SD, 8.1) visits and 0.4 (SD, 4.8) visits in the lonely death group and control group, respectively, in the year 1 year prior to death (P < 0.001) (Fig. 1, Supplementary Table 2). Healthcare utilisation during the 5 years and 1 year prior to death. (A) Outpatient visits in the past 5 years. (B) Outpatient visits in the past 1 year. (C) Inpatient visits in the past 5 years. (D) Inpatient visits in the past 1 year. (E) Emergency visits in the past 5 years. (F) Emergency visits in the past 1 year. In logistic regression analysis of factors associated with lonely death, low income level was strongly associated with lonely death (aOR, 95% CI for the 5th [lowest, MA] income group: 14.188, 12.003–16.772, compared with the highest group). Also, CCI demonstrated a significant association with lonely death. The aOR (95% CI) for a CCI score of 1–2 was 1.105 (1.000–1.220), whereas it was 1.654 (1.429–1.915) for CCI score of ≥ 3, compared with a CCI score of 0. Moreover, specific medical conditions were found to substantially influence the likelihood of lonely death, including diabetes (aOR, 1.436; 95% CI, 1.285–1.606), heart failure (aOR, 2.008; 95% CI, 1.684–2.394), hip fracture (aOR, 2.703; 95% CI, 1.727–4.231), cirrhosis (aOR, 3.012; 95% CI, 2.291–3.960), schizophrenia (aOR, 2.384; 95% CI, 1.738–3.270), bipolar disorder (aOR, 2.131; 95% CI, 1.659–2.738), and alcohol-related diseases, particularly mental and behavioural disorders associated with alcohol use (aOR, 5.482; 95% CI, 4.397–6.836) (Fig. 2). Factors associated with lonely death. CI = confidence interval, MA = medical aid. When analysing the factors associated with lonely death by sex, similar results were observed, but the magnitude of associations differed. The aOR for the 5th income level was greater in males (17.945; 95% CI, 14.751–21.830) than in females (5.995; 95% CI, 4.294–8.368). Although males generally exhibited a higher likelihood of lonely death than females across most conditions, females were more likely to die lonely than males in cases of cirrhosis, schizophrenia, depression, bipolar disorder, and alcohol-related conditions (Fig. 3). Factors associated with lonely death according to sex. CI = confidence interval, MA = medical aid. DISCUSSION In this nationwide study, individuals who experienced lonely death in Korea in 2021 were more likely to be covered under MA, have a low income, reside in urban area (Seoul or metropolitan cities), and have a high CCI compared with age- and sex-matched controls. Healthcare utilisation and various underlying diseases were also more frequent in lonely death cases. Most of these factors exhibited statistical or borderline significance after adjusting for confounders in logistic regression analyses. Stratified analyses by sex demonstrated results similar to those of the main analysis, whereas income level exhibited a stronger association in males than in females. Evidence from previous research on the differences between lonely death and general death is scarce. A study conducted in Los Angeles, United States, where a broader definition of unclaimed bodies was used to define lonely deaths, reported an association of unemployed status among males and low income among females with higher lonely death rates, which is consistent with our results.14 Also, being divorced, never married, or living without family were associated with lonely death rates in the previous study.14 These findings imply that social isolation may result in lonely death, with a socioeconomically vulnerable population being prone to lonely death. In our study, a high CCI score was associated with lonely death only in the lowest-income group after adjustment. Multi-morbidity, represented by a high CCI, increases the risk of adverse health outcomes, including mortality.1516 Furthermore, most of the major diseases were more common in individuals who experienced lonely death than in controls. Chronic diseases such as hypertension and diabetes; mental disorders, including schizophrenia, depression, and bipolar disorder; alcohol-related diseases; liver cirrhosis potentially attributable to alcohol; and diseases which may be associated with immobility, including heart failure, stroke, and hip fracture, were associated with lonely deaths in the adjusted analyses. The number of days of healthcare utilisation during the last 5 years, including outpatient visits, admissions, and emergency room visits, was also higher in the lonely death group than in the control group. Similarly, a previous meta-analysis and systematic review reported an association between social isolation and adverse health outcomes, including cardiovascular diseases and mental health problems.17 Furthermore, multi-morbidity was associated with loneliness and social exclusion in several studies.1819 Notably, one population-based study reported that an increase in mortality according to multi-morbidity is moderated by the level of social support and that multi-morbidity is not associated with mortality in those with high levels of social support.15 Individuals with lower incomes are likely to be more socially isolated and vulnerable to the physical or social risks due to multi-morbidity, which can be associated with lonely death. Therefore, targeted strategies for individuals with low income and multi-morbidity, who may be a very high-risk group for lonely deaths, may be beneficial to reduce lonely deaths. In stratified analyses by sex, diabetes, heart failure, cirrhosis, schizophrenia, bipolar disorder, and alcohol-related diseases were associated with lonely death after adjustment for both sexes, as in the main analysis, whereas schizophrenia, bipolar disorder, and alcohol-related diseases exhibited stronger associations in females. A previous survey in Korea reported higher lonely deaths among males than in females, accounting for 83% of the total in 2021.5 Therefore, strategies for screening and monitoring high-risk groups might be more focused on males. However, our study revealed a high risk of lonely death in females with various underlying diseases, particularly mental disorders and alcohol-related diseases. Notably, alcohol use and other alcohol-related problems, along with mental disorders, are the potential risk factors for suicide.2021 Previous meta-analyses reported a two- and eight-fold increased risk of suicide due to alcohol use and mental disorders, respectively.2021 Although the cause of death was not analysed in our study, death by suicide accounts for a significant portion of lonely deaths in Korea, particularly in young individuals.5 Women with mental disorders or alcohol-related diseases should be monitored more closely to prevent lonely deaths, including suicide. Recent social survey results from the Statistics Korea indicate a continuous decrease in the proportion of people who believe families should provide for their parents, as well as in the number of individuals reporting that they have others to rely on when they need.22 Meanwhile, the proportion of single households continues to increase.5 The conventional caregiving role of the family and the support from an unofficial social network seem to be weakening.12 These findings imply that families or neighbours can no longer function solely as a safety net, thereby highlighting the need for systematic efforts at the local or national government levels. Considering our finding that individuals with poor socioeconomic status are at higher risk of lonely death, along with a previous survey showing that the majority of individuals who experienced lonely death had a history of receiving social security benefits,5 screening vulnerable populations based on the social welfare system may help identify high-risk groups. Notably, careful monitoring is required for males with low incomes. However, limitations may exist in identifying high-risk groups for lonely deaths based on the social welfare system. The proportion of Korean individuals who experienced lonely death but were not identified based on either the MA or the Welfare Blind Spot Identification System, which was implemented in 2015 by the Ministry of Health and Welfare to primarily detect and support the economically vulnerable population, reached 40% in 2021.1223 Our findings, showing associations of multi-morbidity, various underlying diseases, and healthcare utilisation with lonely death, imply that the healthcare system can serve as a platform for the identification of individuals at risk of lonely death who are difficult to identify using other methods. Primary care physicians should collaborate with local governments to contribute to the detection and management of these individuals. Additionally, the development of a systematic screening tool that reflects trends in healthcare utilisation and underlying diseases may be beneficial in the long term. Our study has some limitations. We could not confirm the causality owing to the cross-sectional study design. In addition, we could not collect information related to lonely deaths, such as household type or social relationships, as we used administrative data. Nevertheless, we investigated various characteristics, including underlying diseases and healthcare utilisation, which have not been examined in previous studies. Our study had several strengths. We used a nationwide dataset of all lonely deaths that occurred in 2021 in Korea to ensure representativeness. In addition, unlike a few previous studies that mostly involved individuals who experienced lonely death, we compared their characteristics with age- and sex-matched controls to clarify the possible predictors of lonely death. Using a large sample from a national database, we could perform various stratified analyses while adjusting for multiple confounders to identify high-risk groups. In conclusion, in Korea, individuals who experienced lonely death were more likely to be covered under MA, have a low income, and have multi-morbidity compared with age- and sex-matched controls. Various major diseases, including chronic diseases, mental disorders, and alcohol-related diseases, were associated with lonely death. Furthermore, the number of days of healthcare utilisation was also higher in individuals who experienced lonely deaths than in controls. Targeted strategies for monitoring and managing potentially high-risk groups may help prevent lonely deaths.