To collect data on attempted suicides, we used the National Emergency Department Information System (NEDIS) database in South Korea from January 2016 to December 2020. NEDIS is a nationwide Emergency Department (ED)-based database developed by the National Emergency Medical Center under the Ministry of Health and Welfare. This database collects real-time clinical and administrative data on patients who have visited EDs, including demographic information (age and sex), insurance type, intentionality, injury mechanism, injury severity, geographic location of the ED, route of ED visit, and outcomes (discharge, transfer, and death) [20,21].

Suicide death data between January 2016 and December 2020 were obtained from the MicroData Integrated Service (MDIS) managed by Statistics Korea [22]. We extracted the monthly number of suicide deaths stratified by sex and age from the MDIS.

Patients who visited the EDs after a suicide attempt between January 2016 and December 2020 were included in this study. ED visits for suicide attempts were defined as any ED visit for injury related to intentional self-harm according to the Columbia Classification Algorithm of Suicide Assessment [23]. Suicide deaths were identified by cause of death codes using the International Classification of Diseases, 10th revision codes X60-X84 (Intentional self-harm). Individuals under 6 years of age were excluded from this study.

We used an interrupted time series (ITS) study design, which is widely used to investigate the association between suicide-related outcomes and COVID-19 pandemic [8,10,24]. Using a quasi-experimental design, the ITS analysis was used to assess the impact of population-level health interventions [25]. To evaluate the temporal impact of the COVID-19 pandemic, the monthly number of suicide attempts and suicide deaths before (January 2016-February 2020) and during (March-December 2020) the pandemic were compared.

After the first case of COVID-19 was confirmed in South Korea on January 20, 2020, the number of confirmed cases increased rapidly from February 18, 2020 [26]. On February 29, 2020, the Korea Centers for Disease Control and Prevention Agency implemented a social distancing policy to limit the spread of COVID-19 [27]. Thus, March 1, 2020, was considered the beginning of the COVID-19 pandemic period in the ITS analysis.

The primary outcome was the incidence of suicide attempts or suicide deaths. We stratified the monthly number of suicide attempts according to sex and age group or insurance type. For suicide deaths, we stratified the monthly number of suicide deaths according to sex and age group, occupation, or education level. Age was categorized into the following four groups: adolescents (<18), young adults (18-29), middle-aged (30-59), and older adults (≥60). Insurance type was categorized into five levels, including national health insurance (NHI), medical aid (medical aid type 1 and 2), self-pay, others, or unknown. Occupation was categorized into occupied (8 types of occupation, including managers, professionals, clerks, etc.) and unoccupied (unoccupied, housewife, and others). Education level was classified into four categories, including middle school or less, high school, college or higher, or unknown. Subjects with unknown insurance type, occupation, or education level were not included in the ITS analyses.

In the ITS analysis, we included a continuous variable that counted the number of months from January 2016 as our time variable, a binary dummy variable indicating exposure to the COVID-19 pandemic (0: before the pandemic, 1: during the pandemic), and a continuous variable that counted the number of months after the emergence of COVID-19 (0: before the pandemic, time-50: after the pandemic). Additional covariates were the Fourier terms (pairs of sine and cosine functions) used to adjust for seasonality.

The annual incidence rates per 100,000 person-years were calculated using the mid-year census population of the respective years in Korean Statistical Information Service (KOSIS). We calculated the age-and sex-specific incidence rates of attempted suicides and suicide deaths and their corresponding 95% confidence intervals (CI) for 2019 and 2020. We then calculated the incidence rate ratios and corresponding 95% CI to compare the incidence rates in 2019 and 2020.

ITS analysis was performed to identify changes in the monthly number of suicide attempts and suicide deaths before and during the COVID-19 outbreak. The pandemic onset was defined as March 2020. To handle overdispersion, we used quasi-Poisson regression and quantified the relative risk (RR) and 95% CI. A time variable to indicate the pre-pandemic trend, a binary indicator variable estimating the level change in the number of suicide attempts or suicide deaths, and a continuous variable that counted the number of months during the COVID-19 pandemic to measure the changes in the slope during the pandemic were included in the model. We used Fourier terms to adjust for seasonality. After fitting the model, residual analyses were conducted to test autocorrelation.

We extended the quasi-Poisson regression model to forecast the monthly number of suicide deaths. Using data from January 2016 to December 2020, we forecasted the monthly number of suicide deaths and calculated the corresponding 95% prediction interval for the period from January 2021 to July 2022 based on the forecasting scenarios. These scenarios are as follows: 1) no further effects of the COVID-19 pandemic, 2) continuous effects of the COVID-19 pandemic, and 3) based on the periods before and during the COVID-19 pandemic [24]. To validate our forecasting scenarios, we compared the forecasted and actual estimated monthly number of suicide deaths. The actual estimated number of suicide deaths was calculated by fitting a quasi-Poisson regression model using suicide death data obtained from KOSIS between January 2016 and July 2022.

All analyses were performed using R programming language version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria). A two-tailed p-value of <0.05 was considered statistically significant.

This study was performed in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of the Yonsei University (4-2021-0259). Informed consent from the participants was waived by the Institutional Review Board of the Yonsei University due to the retrospective nature of this study.

The changes in monthly attempted suicide patterns during the COVID-19 pandemic according to sex and age groups are shown in Table 3, Figure 1, Supplementary Table 3 and Supplementary Figure 1. Until February 2020, the total number of suicide attempts increased (RR [95% CI] pre-pandemic trend: 1.01 [1.01-1.01]). After the onset of the pandemic (March 2020), the number of suicide attempts immediately decreased, but there was no significant change in the slope compared with the pre-pandemic trend (RR [95% CI] level change: 0.88 [0.80-0.97], slope change: 0.99 [0.98-1.01]).

Stratified by sex and age, the results from the ITS analyses showed that the immediate reduction of attempted suicides was significant only among males (RR [95% CI] level change: 0.85 [0.76-0.94]), and not among females (RR [95% CI] level change: 0.90 [0.82-1.01]) after the pandemic began. Specifically, the monthly number of suicide attempts immediately dropped in adolescent boys (RR [95% CI] level change: 0.65 [0.50-0.87]) and girls (RR [95% CI] level change: 0.56 [0.40-0.78]). For older males, the number of suicide attempts declined sharply at the beginning of the pandemic but thereafter, the slope increased by 3.0% against the pre-pandemic levels (RR [95% CI] level change: 0.67 [0.59-0.77], slope change: 1.03 [1.01-1.05]). However, the slope of suicide attempts among older females did not change during the pandemic compared with the pre-pandemic period (RR [95% CI] level change: 0.83 [0.71-0.97], slope change: 1.01 [0.99-1.03]).

Stratified by insurance type, the monthly number of suicide attempts shortly declined in patients with NHI (RR [95% CI] level change: 0.89 [0.80-0.98]), medical aid (RR [95% CI] level change: 0.82 [0.71-0.95]), and self-pay (RR [95% CI] level change: 0.84 [0.71-0.99]). There were no significant changes for patients with other insurance at the beginning of the pandemic. However, there was an increasing trend during the pandemic compared to the pre-pandemic trend (RR [95% CI] level change: 1.20 [0.77-1.88], slope change: 0.90 [0.83-0.97]) in patients with other insurance.

The results of the association between the COVID-19 pandemic and changes in the number of suicide deaths are shown in Table 4, Figure 2, Supplementary Table 4 and Supplementary Figure 2. The total number of suicide deaths did not show specific changes during the COVID-19 pandemic (RR [95% CI] level change: 0.95 [0.85-1.07], slope change: 1.00 [0.97-1.01]). In addition, no statistically significant changes were observed in any of the subpopulations except older males. The number of suicide deaths significantly decreased at the beginning of the pandemic, but the trend did not change during the pandemic in older males (RR [95% CI] level change: 0.76 [0.66-0.88], slope change: 1.02 [1.00-1.04]).

The forecasted and validated suicide death trends by sex between January 2021 and July 2022 are shown in Supplementary Tables 5-7 and Supplementary Figure 3.

This sudden decline in suicide attempts among adolescents can be attributed to nationwide school closures and the implementation of online classes in South Korea. This is in line with previous studies examining the mental health of adolescents in South Korea that showed that suicidal ideation and attempts were lower initially in the pandemic (2020) than in previous years; these suggest a close association with school closure and in-home online classes by lowering the burden of school activity, academic achievement, stress from interpersonal relationships in school which could have been risk factors for suicide attempts in adolescents [28,29].

In our sex-specific analysis, we observed a heterogeneous impact of the pandemic on suicide attempts among adolescents. Compared with adolescent boys, although the increasing trend during the pre-pandemic period was more evident in adolescent girls, the decline in the level change was more pronounced among girls. These findings suggest that the pandemic had a more significant impact on girls than boys, which can be resulted from the differences in risk factors for suicide attempts in adolescents based on sex [30]. Given the long-term implications of suicide attempts in younger generations [31], further investigation will be needed to fully account for the sex differences in suicide attempts during the pandemic.

Our study observed initial reductions in suicide attempts among older adults and in suicide deaths among older males. These declines may be explained by a phenomenon called the “honeymoon effect” [32]. The increased sense of belonging and empathy during the early phase of the epidemic could be contributed to short-term decreases in the risk of suicide-related outcomes [33,34]. Conversely, the decline in suicide attempts could be associated with decreased help-seeking behaviors in other adults. The excessive fear of becoming infected with COVID-19 due to the higher mortality risk [35] might lead to reduced help-seeking behaviors, especially among older adults. In line with this, the number of ED visits by older adults during the early phase of the pandemic also declined [36-38].

Following the immediate decline at the initial stage of the pandemic, suicide attempts and suicide deaths among older males rebounded during the pandemic compared to the pre-pandemic years, although the increase in suicide deaths was marginally significant. Older adults are more susceptible to social isolation, and the effect of social isolation is generally greater in males than in females [39,40]. Physical distancing measures during the COVID-19 pandemic have reduced social participation and access to community services over time, negatively impacting the overall well-being of the older population [41]. Moreover, since older adults are vulnerable during the disaster, feelings of being a burden to families and younger generations could exacerbate the suicide risk [42]. As the pandemic continues, feelings of loneliness and isolation, and perceived burdensomeness will have a greater impact on the mental health of older adults, especially older males. These factors may contribute to an increase in the risk of suicide attempts or suicide deaths.

This study has several limitations. First, since the current study obtained data on suicide attempt incidence from the ED visit data, the observed number could have been underestimated because some individuals may not have sought help after suicide attempts. Second, the collected covariates do not rule out all other confounders that may affect suicide attempts and deaths, including a history of mental illness and individual economic status, which are known risk factors for suicidality and were not assessed in this study. Therefore, further studies are needed to investigate individual-level risk factors that affect suicide-related outcomes during the COVID-19 pandemic. Third, our study only covered the first year of the COVID-19 pandemic, which is relatively short for capturing the long-term effects of the pandemic on suicide attempts and suicide deaths. Given that the COVID-19 pandemic has persisted for several years and may continue to influence the trends of suicide attempts and suicide deaths, future studies with extended time frames will be needed to provide a more comprehensive understanding of the pandemic’s impacts. Finally, this study was conducted on a South Korean population. As the impact of the COVID-19 pandemic on suicidality varies between nations and across time, attention should be paid when applying our results to other countries.

In conclusion, our study provides a comprehensive overview of the multifaceted impact of the COVID-19 pandemic on suicide attempts and suicide deaths using nationwide data. Our findings revealed that the effects of the pandemic on these outcomes were disproportionate, with variations observed across sex and age groups, specifically among adolescents and older adults. These results highlight the importance of considering sex and age differences when formulating and implementing targeted suicide prevention measures in the aftermath of a pandemic.