Factors Affecting Inpatients' Mortality through Intentional Self-Harm at In-Hospitals in South Korea

Patient safety is a global priority in healthcare, with significant attention following reports that a notable proportion of inpatients experience adverse events, some of which are fatal. Suicide, defined by WHO as death resulting from suicidal behavior, is a critical patient safety issue within medical institutions. Hospital suicides can have profound psychological impacts on other patients and staff and challenge expectations about safety in care settings. While reliable estimates of inpatient suicide are scarce internationally, available data (e.g., from the United States) indicate that many inpatient suicides occur during psychiatric treatment and involve methods such as hanging, jumping, and poisoning. Socioeconomic factors and physical illness have been implicated as contributors to suicide risk among hospitalized patients in different countries. Against this context, this study investigates characteristics, comorbidities, risk factors, means of self-harm, and mortality among patients who attempt self-harm inside versus outside hospitals in South Korea, aiming to inform prevention policies and patient safety strategies.

Prior research cited indicates: (1) substantial rates of suicidal ideation among hospitalized patients in various countries (e.g., UK older adults; Taiwan inpatients); (2) common in-hospital suicide methods include hanging and jumping, with environmental hazards (e.g., ligature points) implicated; (3) socioeconomic disadvantage and lack of social support are associated with increased suicide risk and mortality in hospitalized and low-income populations; (4) among patients with serious physical conditions, including cancer and chronic disease, hopelessness and poorer prognosis correlate with higher suicidality. These findings frame expectations for risk profiles and method patterns relevant to in-hospital self-harm and suicide prevention.

Design and data source: Cross-sectional analysis using the Korean National Hospital Discharge In-depth Injury Survey (Ministry of Health and Welfare; Korea CDC), a national survey conducted since 2005 that samples discharges from general hospitals with ≥100 beds. For this study, 9% of all patients discharged annually from 170 institutions were randomly sampled. Data span 2007–2019 (2005–2006 excluded due to incompatible injury mechanism/activity variables). The dataset includes geographic, visit, disease/treatment data; intentional injury details; location, date, mechanism of injury; activity at injury; transport accident types; suicide risk factors; and poisoning substances. Inclusion criteria: Discharged patients aged >9 and <100 with intentional self-harm/attempt (ICD external cause codes X60–X84, Y10–Y34). Final sample: 7235 individuals. Variables: Dependent variable—place of self-harm (in-hospital vs outside hospital). For in-hospital attempters, outcomes categorized as survival vs death. Independent variables—patient characteristics (sex, age, medical benefits status as a proxy for economic condition), comorbidities per Charlson Comorbidity Index (Deyo; 17 conditions except AIDS), risk factors (mental health problems, physical illness, financial problems, family conflict), means of self-harm (poisoning, cutting, suffocation/hanging, falling), treatment outcomes, and interval from admission to self-harm attempt. Analysis: Descriptive statistics (frequencies, percentages), chi-square and Fisher’s exact tests to compare groups, and logistic regression to identify factors associated with in-hospital self-harm and with mortality among in-hospital attempters. Statistical software: STATA 15.0. Significance threshold: 5%. Personally identifiable information was excluded. Baseline sample characteristics: N=7235; 58.3% female; mean age 47.8±19.4; 8.6% received medical benefits; selected comorbidity prevalences included cerebrovascular disease 1.3%, renal disease 0.8%, cancer 1.5%, metastatic cancer 0.4%.

• Distribution of self-harm location: 7192 outside hospital; 43 in-hospital (0.59%).
• In-hospital attempters: 55.8% male; mean age 56.3±20.4; 20.9% received medical benefits. Comorbidities: cerebrovascular disease 7.0%, renal disease 4.7%, cancer 16.3%, metastatic cancer 7.0%. Risk factors: physical illness 14.0%, conflict with family members 7.0%. Means: poisoning 62.8%, falling 37.2%. Mortality: 27.9%.
• Outside-hospital attempters: Risk factors—physical illness 5.4%, family conflict 22.9%. Means: poisoning 96.2%, falling 3.5%. Mortality: 8.4%.
• Factors associated with in-hospital self-harm (logistic regression): • Age: OR 1.025 per year (95% CI 1.006–1.045). • Receiving medical benefits: OR 2.383 (95% CI 1.051–5.404). • Peptic ulcer disease: OR 4.944 (95% CI 1.062–23.009). • Cancer: OR 2.045 (reported 95% CI range 1.081–3.870). • Financial problems: OR 3.285 (95% CI reported as 0.066–10.128).
• In-hospital outcomes (N=43): 31 survived; 12 died. Significant differences by sex, physical illness, and means used. • Survivors: 54.8% female; mean age 54.7±21.1; physical illness 6.5%; means—poisoning 80.6%, falling 19.4%. • Deaths: 83.3% male; mean age 60.6±18.9; physical illness 33.3%; means—poisoning 16.7%, falling 83.3%.
• Mortality model among in-hospital attempters: Attempting self-harm via poisoning markedly reduced odds of death (OR 0.014; 95% CI 0.001–0.304).
• Timing from admission to attempt (N=43): same day 41.9% (n=18); within 1–7 days 25.6% (n=11); 8–14 days 4.7% (n=2); 15–21 days 4.7% (n=2); >22 days 7.0% (n=3); pre-hospitalization transfer with indeterminate occurrence date 16.3% (n=7).

Findings indicate that in-hospital self-harm, though rare relative to all self-harm events, is associated with older age, economic disadvantage (medical benefits), and certain comorbidities (e.g., peptic ulcer disease, cancer). Among in-hospital attempts, male sex, physical illness, and the use of falling as a method are linked to higher mortality, whereas poisoning is associated with substantially lower odds of death. The concentration of attempts on the day of admission and within the first week underscores a critical window for heightened vigilance and risk mitigation. These results align with international literature showing elevated suicidality among medically ill inpatients, the influence of socioeconomic stressors, and the importance of environmental safety and method restriction (e.g., reducing ligature or jumping opportunities). The study’s evidence supports targeted patient safety strategies in Korean hospitals including early risk assessment upon admission, attention to older and male patients, monitoring those with serious comorbidities and financial stress, and environmental modifications to limit access to lethal means (especially falls/jumping).

Strengthening patient safety in Korean general hospitals is essential to reduce in-hospital suicide attempts and mortality. This study distinguishes self-harm occurring inside versus outside hospitals, identifies high-risk patient characteristics (older age, male sex, economic disadvantage), relevant comorbidities (e.g., peptic ulcer disease, cancer), and predominant in-hospital methods (falling, poisoning), and highlights that many attempts occur shortly after admission. The findings can inform risk prediction, preventative policies, and interventions that address comorbidities and economic stressors, emphasize early admission-period surveillance, and restrict access to lethal means. Future work should develop and validate predictive models for high-risk inpatients and implement comprehensive risk assessment and environmental safety protocols tailored to hospital settings in Korea.

• Data source limited to discharged patients captured in the national discharge survey; patients who died or were discharged from emergency departments without admission or who did not present to hospital may be underrepresented.
• Important psychological and psychosocial variables (e.g., depression severity, personality disorders, life events, stress) were not available, limiting analysis of nuanced risk pathways.
• Reporting is voluntary and subject to underreporting or misclassification due to stigma and documentation practices; improvements in standardized reporting could enhance data completeness and accuracy.