The findings of this study demonstrated that the methods of suicide attempt had an effect on the need for hospitalization. Among the developed countries, South Korea has the highest suicide rate, which is two times higher than that in the United States . The ED is an important point-of-care in the healthcare process for suicide attempters. In this study, drug overdose was the commonest method of suicide (52.4% of the total patients), and had similar rates in the hospitalization and discharge groups. Walker et al had reported drug overdose to be a strong predictor of ICU admission  | Kim et al classified stabbing, hanging, drowning, and jumping off a height as clinically serious, lethal methods, and severe depression, psychological disturbance, and repetitive suicidal ideation as factors that indicated high medical severity . This study excluded patients who died while in the ED or within 24 h of hospitalization, and those patients who represented less than 10% of all suicide attempters.
Salles et al reported that an alcohol use disorder was unassociated with hospitalization for inpatient in psychiatric care, whereas depression clearly was an associated factor . This study showed that alcohol use was unassociated with hospitalization; rather, alcohol use was higher in the discharge group. However, this study only investigated whether the patients consumed alcohol instead of identifying if the patients were diagnosed with an alcohol use disorder, and the amount of alcohol consumed and the blood alcohol concentration was not ascertained. These aspects need to be evaluated in future research.
Psychiatric patients constitute one in eight patients who visit the ED, contributing to ED overcrowding [21, 22]]. The ED LOS is associated with increased ED crowding. The walk-out rate was 0.34 patients/hour when there were no patients staying in the ED for less than 6 hours and 0.77 patients per hour when the patients’ ED LOS was more than 6 h . In this study, the average ED LOS in the hospitalization group was 11.4 ± 18.8 h vs 4.2 ± 12.3 h in the discharge group, which significantly contributed to the ED crowding. The overall ED LOS of patients with psychiatric emergencies and suicidal/self-harm patients showed a median duration of 2.4 h, which shows that the duration has been maintained at a similar value for 3 years, with local differences . In this study, the hospitalization group had a significantly longer ED LOS than the discharge group, with the average duration being well over 6 h. Besides the time for assessment and treatment, the time to take hospitalization decision contributed to the long ED LOS. Both of the hospitals where this study was conducted undertook the hospitalization of suicide attempters via drug overdose, hanging, and drowning in the ED. South Korea does not have a dedicated toxicology department at hospitals, leaving the hospitalization decision to the ED or internal medicine physicians. Therefore, the ED LOS is extended when the patients cannot be admitted to a psychiatric ward due to physical injuries and is even longer with ED crowding.
In this study, the ORs for hospitalization based on mental status at presentation were 2.027 for verbal response, 6.200 for pain response, 39.931 for unresponsiveness. The OR for consciousness was 2.330, indicating that emergency physicians should not delay the hospitalization decision when the patients were not alert, in order to avoid ED crowding and improve the ED occupancy ratio. Jo et al reported that ED crowding was associated with a higher mortality rate in critically ill patients . In addition to the level of alertness, the socioeconomic status can affect the hospitalization decision. Even in alert patients, a report estimates that approximately 25% of patients could have been discharged if they had social support, and that clinical severity alone does not determine the need for hospitalization . In this study, consultancy care was provided to the caregivers when hospitalized patients had altered mental status, and unconscious patients who received a psychiatric consultation after they regained consciousness. All hospitalized patients were transferred or discharged after the psychiatric follow-up consultation.
The RRRS had high sensitivity and specificity for the hospitalization decision of suicide attempters in this study. The RRRS was validated for psychometric property  and has been used to determine the lethality of suicide attempts . Oh et al showed that deliberate self-poisoning was included in the high-risk/low-rescue group . Kim et al analyzed several psychologic scales that are used to determine the hospitalization need of suicidal patients in the ED and reported that RRRS was the most useful factor for predicting hospitalization in the ED setting. Baca-García et al reported that psychiatrists appear to rely on patients' self-report to decide on hospitalization rather than demographic, diagnostic, or psychosocial factors . A few studies have suggested previously that suicide mortality can be lowered if individuals at risk of suicide are effectively identified and appropriately treated. Emergency physicians evaluate and treat suicide attempters and make decisions to admit patients to the medical ward and not to the psychiatric ward. Weiland et al reported that emergency physicians were uniformly confident in their decision to shift patients at risk of suicide or self-harm to the inpatient department . In the busy ED, physicians do not consider scoring systems. Emergency physicians have limited time and high patient volumes . Therefore, a guideline for early decision-making for hospitalization of suicide attempters would be helpful to avoid ED crowding.
In the RRRS, the absence of loss of consciousness, confusion, and coma are assigned 1, 2, and 3 points, respectively. However, it does not have a large effect on the total score. Based on a regression analysis, the level of consciousness should be considered as a single factor for the hospitalization decision.
Planned suicide attempt was twice as frequent in the hospitalization group than in the discharge group (OR 1.728). A previous report indicated that planned suicidal attempts have more severe medical consequences for serious suicidal attempts . For patients with planned suicidal attempt, a psychiatric follow-up after initial care is especially important.
In this study, the history of psychiatric care or current psychiatric medication usage did not differ significantly between the study groups. This may suggest that suicidal behavior was present regardless of psychiatric treatment, or that psychiatric treatment was inadequate to prevent suicidal attempts. Harada et al reported that females tended to be over-represented in the psychiatric consultation group, and males in the non-consultation group. Poisoning by prescription drugs was used more frequently as a method of suicide in the consultation group. Moreover, the prevalence of adult personality disorders and schizophrenia and related disorders were higher in the consultation group than in the non-consultation group .
Shepard et al reported that the national cost of suicides and suicidal attempts in the United States in 2013 was $58.4 billion, based on the reported numbers alone. However, after adjustment for under-reporting, the cost increased to a total of $93.5 billion, which was 2.1–2.8 times that reported in previous studies . In South Korea, statistics on suicidal attempts are only available through ED data. Furthermore, the cost of medical care and related indirect costs have not been investigated. Considering the high prevalence of suicide in South Korea, if the direct and indirect costs of ED use by suicide attempters and by under-reporting of suicide attempts are studied, the national cost of suicide is likely to be very high. Further research is needed to assess the national costs associated with suicide.
The limitations of this study include the fact that many patients refused data collection or were excluded due to missing data. Moreover, the study was conducted at tertiary teaching hospitals in metropolitan areas, and may not be representative of ED situations in more rural areas. Further research should include multiple levels of hospitals nationwide to identify other issues in the medical hospitalization process of suicide attempters.
In this study, patients in the ICU and GW were grouped together in the hospitalization group and compared with those in the ED discharge group. In both hospitals where this study took place, patients who attempted suicide by drug overdose, hanging, and drowning were hospitalized from the ED, and patients were admitted to the ICU if they had a high risk of suicide re-attempt even if their physical injury was treatable in the GW, as close observation is available in the ICU. Therefore, the hospitalization rate for the ICU and GW was difficult to determine, and it was more appropriate to include both as the total hospitalization group. Furthermore, continuous psychiatric consultation is not provided during hospitalization, and is only provided following discharge. Another limitation is that bed availability at the time of hospitalization was not assessed. It may be possible that an early hospitalization decision was made, but a bed was not immediately available. However, this may not add a significant bias as patients are typically transferred to another hospital when the bed is unavailable soon after the hospitalization decision.
This study included 2.6% of ED visits by patients who were 19 years of age or older without an existing illness in South Korea. When suicidal/self-harming patients desired medical hospitalization due to physical injury, there is an extra barrier to hospitalization decision for departments other than psychiatry due to the patients’ psychiatric conditions combined with their physical injury. The present study demonstrates that the level of consciousness and RRRS scores can be used as indicators for the hospitalization in suicide attempters.