Abstract
Background The coronavirus disease (COVID-19) pandemic is an unprecedented public health crisis, but its effect on suicide deaths is little understood.
Method We analysed data from monthly suicide statistics between January 2017 and September 2020 and from online surveys on mental health filled out by the general population during the COVID-19 pandemic in Japan.
Outcomes Compared to the last three years (2017-2019), the number of suicide deaths was lower than average during the initial phase of the pandemic but, starting in July 2020, exceeded the past trend. The source of the increase was mainly female suicides, whose numbers increased by more than 20% in August and September 2020. Suicides by young women (less than 40 years of age) experienced a 63.1% increase in August 2020 compared to the same month in the past three years. The number of suicides among students was also notably higher in August and September 2020 than the corresponding numbers of past years. Our survey data indicated the status of mental health among young women was worse than that of women above 40 years and also that young female workers were more likely than any other group to have experienced a job or income loss in recent months, suggesting adverse economic conditions surrounding some of these individuals.
Interpretation Our results strongly indicate that continuous monitoring of mental health, particularly that of the most vulnerable populations identified in this study, and appropriate suicide prevention efforts are necessary during, and in the aftermath of, the COVID-19 pandemic.
Introduction
The coronavirus disease (COVID-19) pandemic is an unprecedented public health crisis that entails both physical and mental health consequences. Emerging evidence suggests that the COVID-19 pandemic severely affects mental health in the general population.1,2 These findings correspond with the concerns that deteriorated mental health, in combination with higher unemployment rates during the pandemic, have the potential to increase the incidence of suicide worldwide.3,4
Despite this possibility, to our knowledge, no peer-reviewed study has documented how the current pandemic has affected the prevalence of suicide at the national-level by major demographic group, due to the lack of reliable fast-reporting data. In response to this, this study examined the suicide deaths and mental health status of the general population during the COVID-19 pandemic in Japan. We focused on Japan for two reasons. First, Japan has one of the highest suicide rates in the OECD countries.5 Its suicide rate (the number of suicide deaths per 100,000 population) as of 2018 was 16.5; more than 20,000 people ended their life by suicide in that year alone.6 Thus, the vulnerable population in Japan may be highly responsive to the crisis, which could result in a rapid increase in suicide deaths. Second, rapid-release monthly suicide statistics, in combination with our own monthly survey data, would allow us to closely monitor the status of the general population since the onset of the pandemic.
Japan had its first encounter with COVID-19 relatively early compared to the rest of the world. The first case was reported on January 16, 2020, followed by outbreaks on a cruise ship at Yokohama Port and in Hokkaido in February 2020. The first wave of COVID-19 cases was observed in April 2020, and a second, larger wave of cases started in July 2020. As of August 31, 2020, the total number of confirmed COVID-19 cases was 67,077, and the number of deaths attributed to COVID-19 was 1,278, which translates to 10.15 deaths per one million persons. The number of new COVID-19 cases in Japan is shown in Fig 1.7
Note: The shaded area indicates the period of the state of emergency (April 7-May 25).
The Japanese government has imposed several measures to stem the tide of the pandemic. School closure started in early March. The Japanese government declared a state of emergency in major metropolitan areas on April 7, 2020, which was extended to the rest of Japan on April 16. Without introducing lockdown measures or strict domestic movement restrictions during the state of emergency, authorities requested non-essential businesses to close or opt to work remotely. Stores and restaurants were asked to operate for reduced hours. The state of emergency was lifted on May 25, 2020.
Methods
We analysed the publicly available monthly suicide statistics tabulated by the National Police Agency (NPA) and published by the NPA and the Ministry of Health, Labour and Welfare of the Japanese government.8,9 The NPA suicide statistics are based on police investigations of suicide cases. In Japan, the police investigate all suspected cases of suicide, and the agency records all deaths that are determined to be suicide in its suicide statistics data. We used its monthly tabulation from January 2017 and the most recent available data, September 2020. There are two types of NPA data: the data tabulated based on the date that the deceased was found and the data tabulated by the date of death. We used the former, as this is the most frequently updated data.
The NPA suicide data constitute one of the two official data sources on suicide deaths in Japan.6 Another data source is the Vital Statistics of Japan that are drawn from the death certificates issued by physicians. The number of suicide deaths reported in these two data sets are closely related; For example, the number of suicides in 2018 according to the Vital Statistics was 20,465, and the number according to the NPA data was 20,840.6 This study analysed the NPA data because the monthly tabulation of the Vital Statistics data for early 2020 would not be released to the public until late 2020.
We examined the total number of deaths by suicide, the number of suicides by sex, age group (less than 40 years old, 40-59 years old, and 60 years and older), and major occupation. We calculated monthly averages between 2017 and 2019 and compared them with the number of suicide deaths in the corresponding months in 2020 by calculating the percentage changes between the two periods. We also conducted regression-based analysis using the same data (Supplementary File 2). This part of the analysis used publicly available data; thus, no ethics approval was necessary.
In addition, we assessed the status of mental health among the general Japanese population by using data from a series of monthly online surveys that we had been conducting since April 2020. We selected the respondents from a commercial online panel so that they were representative of the Japanese population in terms of their sex, age group, and area of residence. Each monthly survey contained 1,000 respondents. A more detailed description of the survey is provided in Supplementary File 1. For the purposes of this study, we used the surveys taken between April and September 2020 (N=6,000).
As measures of mental health, we analysed the prevalence of depressive and anxiety symptoms, measured by the Patient Health Questionnaire nine-item scale10 and the seven-item General Anxiety Disorder scale (GAD-7),11 respectively. We also examined the self-reported change in employment status of the respondents in the labour market to understand whether they had experienced major job-related changes during the COVID-19 pandemic. We only asked about the changes in employment status since June, and we used the data between June and September 2020 when we analysed the job-related changes of the respondents (N=4,000). We calculated the presence of depressive symptoms (PHQ-9 ≥ 10) and anxiety symptoms (GAD-7 ≥ 10) for all respondents, as well as the percentages of those in the labour force who had lost their job, taken leave or been laid off from work, or experienced significant reduction in working hours in the last three months, and stratified them by sex and age group.
The survey was approved by the Ethics Review Committee on Human Research of Waseda University (approval #: 2020-050) and Osaka School of International Public Policy, Osaka University. We used Python (version 3.7.3) to conduct all tabulations and regression analyses.
Results
The top panel of Fig 2 shows the percentage changes in the number of suicide deaths in Japan overall and by sex. The baseline period is the mean of 2017-2019 for each month. The number of suicide deaths in the two time periods is provided in Supp. Fig 1. The shaded area in the figure indicates the period of the state of emergency. During this period (April and May 2020), Japan experienced approximately a 20% decrease in suicide deaths. However, the declining trend observed in the spring reversed its course in July. By August, the total number of suicides was 7.72% higher than the average number of suicides in the same month of the previous three years. The source of the increase was mainly female suicides; the number of suicides by women in August and September increased from 531.66 and 529 (the mean in 2017-2019) to 651 and 640 (in 2020), respectively, which corresponded to a 22.44% and 20.98% increase.
According to the bottom panels of Fig 2, which report the percentage increases of suicide by sex and age group, suicide deaths by all demographic groups exhibited the same trajectories, with an initial decline during the state of emergency, followed by an increase initiated in July. We observed the largest increase in the summer months among young individuals (less than 40 years old) for both sexes. In particular, the incidence of suicide among young women was 63.1% higher in August 2020 (N = 193) than the average of the same month in the previous three years (mean = 118.33). A similar increase was observed among young women in September, with a 53.7% increase. The observed differences are statistically significant, as shown in Supp. Fig 3.
Fig 3 reports the percentage changes in the number of suicides between the previous three-year averages and the 2020 figures by major occupation categories. There were no suicides by male homemakers in our study period. Similarly, there was a monthly average of fewer than ten suicides by job-seeking women, thus it is not shown in the figure. The trend in suicide deaths by occupation is shown in Supp. Fig 2 and the results of regression-based analysis are reported in Supp. Fig 4.
Fig 3 indicates that students and homemakers experienced the largest increase after the initial decline during the state of emergency. For example, student suicides declined by nearly 25% during the school closure period (early March – late May or early June, depending on the area), then subsequently increased in August and September. Student suicides in August and September averaged 62.66 and 66.66, respectively, during the past three years but increased to 114 and 106 in August and September 2020. Among them, 42 and 46 were university students in August and September, respectively, compared with 24 in each of these months in recent years. The incidence of student suicides was higher among female students. The number of suicides by female students increased by 162.5% and 103.7% in August and September, respectively, compared to the baseline period. The incidence of suicide among homemakers also increased during the summer months, rising by 32.6% in August 2020 (N = 107) compared to the average figure (Mean = 80.66) over the past three years.
Among the employed, we also observed a larger increase of suicides in July and August among females than among males. As reported in the bottom-right panel in Fig 3, the number of suicides by employed women has increased since July 2020. By September, it was 42.86% higher compared to the same months over the last three years.
Fig 4 reports the percentages of those who were categorised as having depressive symptoms or anxiety disorders according to sex and age group. The top panel shows the prevalence of depression based on their PHQ-9 score, and the bottom panel shows the percentages of those who were considered to have anxiety disorders based on their GAD-7 score. Overall, relatively young individuals, those less than 40 years old, were more likely to have depressive and anxiety symptoms than those in other age groups. The June data reported the worst status of mental health for these young individuals, which was right after the state of emergency was lifted.
Finally, Table 1 shows the percentages of survey respondents in the labour force who reported that they had lost their job, taken leave/been laid off from work, or experienced a drastic decrease in working hours over the past three months, stratified by sex and age groups. Overall, female respondents were more likely to have reported changes in their employment status and working hours, with 22.65% of female workers reporting this, compared to 17.07% of male workers. Among female workers, 27.71% of young individuals (less than 40 years old) reported drastic changes in employment and working conditions, whereas the corresponding number for young male workers was lower at 18.32%.
Discussion
This study examined the monthly trajectories of suicide deaths in Japan during the COVID-19 pandemic. We found that the number of suicide deaths was lower during the initial phase of the pandemic in 2020 than during the period of 2017-2019. Starting in July, however, the number exceeded the trend of the past three years. Across all age and occupation groups, the largest increase was found among relatively young women, those less than 40 years old at the time of their death. The prevalence of suicide among students and homemakers in summer months was also notably higher in 2020 than the monthly averages of 2017-2019. These observed differences were statistically significant as shown in Supplementary File 2.
Potential factors behind the large increases in suicide by female and young individuals during the second phase of the pandemic include the economic consequences of the pandemic, school closure, and media reporting of celebrity suicides. The economic damage of the current pandemic was particularly evident in the industries that are served mainly by women, such as the service, retail and travel industries.12 The labour statistics in August 2020 indicate that the number of employed individuals in non-permanent positions, such as part-time or contract workers, decreased for the six consecutive months from the onset of the pandemic. In August alone, compared to the same month in the previous year, the number of these precarious workers decreased by 1.2 million, out of which 0.84 million were women.12 The economic impact of the current pandemic, therefore, was largely felt by women of precarious employment status. It should also be noted that the occupation categories reported in the NPA data are based on the occupation at the time of death, and some women may have been categorised as homemakers if they had lost their job prior to their death.
The results of our survey also revealed the unfavourable psychological and economic conditions of young women. We found that the status of mental health among young women (less than 40 years of age) was worse than that of women above 40 years of age, which may underlie their relatively high suicide rate in the second half of 2020. Relatively worse mental health conditions among young and economically vulnerable individuals were also reported in another study, in which other attributes of the respondents were explicitly taken into account.13 In addition, as reported in Table 1, young female workers were more likely to have experienced a job or income loss in recent months compared to any other groups, suggesting adverse economic conditions surrounding some of these young female workers.
The increase in student suicides might be associated with school closure and an unusual school calendar this year. Schools in Japan closed around March 2, after the former Prime Minister, Shinzo Abe, abruptly announced a request for school closure on February 27, 2020, and they did not reopen until the state of emergency was lifted in late May. Thus, elementary to high school students had to stay at home for almost three months. Emerging evidence indicates the negative impact of school closure on students; a survey conducted between June and July (after school reopening) by the National Center for Child Health and Development reported that 72% of surveyed students (age 7-17 years) reported symptoms that indicated some form of stress reaction.14 Similarly, the mental health condition of students (mainly university students) in our survey was worse than most of the other occupation categories (Supplementary Table 1). Indeed, 28.46% of the surveyed students (N = 123) were classified as depressed.
Returning to school after a summer break is known to be difficult for some students even in normal years, and student suicides have previously peaked on the first day of school after the summer break, specifically on September 1st.15 Many schools reopened in August this year because of the shortened school days in the spring, and some of the increases in August may be attributable to the shifted school calendar.
However, the school calendar cannot explain the high incidence of suicide by university students as they did not return to school in August. Most universities in Japan switched to online teaching when the new academic year started in April, and many continued with this format after the state of emergency was lifted in late May.17 It is possible that some university students were depressed, as indicated by our surveys, because online and solitary learning without on-campus activities largely continued until late September. This is in sharp contrast to students in elementary and high schools, who returned to school by June. It is important to note, however, that the number of student suicides was lower during the school closure than during the past three years, which was also reported in another study,16 and more work is necessary to understand the association between school closure and student suicide.
Media reporting on celebrity suicides might also explain some of the increases in suicide deaths observed in this study. Several well-known individuals died by suicide during the pandemic, and two notable instances were the death of a male actor, who died at the age of 30 on July 18th, 2020, and another suicide death of a 40-year-old actress on September 27th, 2020. Both of them were at the height of their career and the media reporting on their unexpected suicide deaths might have affected vulnerable individuals. The subsequent increase in actual suicides is known to be particularly large when celebrity suicides are widely discussed on SNS or when they are regarded as a “surprise,”18,19 both of which are applicable to the above two cases.
Although the number of suicides started increasing in July 2020, Japan experienced a large decline in suicide deaths at the initial phase of the pandemic when the state of emergency was in place. Two factors may explain the temporary decline. First, it might be attributable to remote work and associated reduced commuting during the period. During the state of emergency, many non-essential businesses switched to remote work, which is a rare practice in Japan, and, consequently, many workers were spared from commuting. In the first week after the declaration of the state of emergency, the number of commuters at major stations in Tokyo during the peak commuting time on weekdays declined to 30% of the pre-COVID levels.20 Remote work may have eased the physical and psychological burdens of a long commute and possibly long working hours. Second, the decline might be explained by enhanced social connectedness during times of crisis. For instance, the level of social connectedness and altruism is known to increase in the aftermath of natural disasters,21 and if natural disasters and the current pandemic share certain similarities, enhanced connectedness may have worked as a protective factor with regard to suicide risks.
This study has several limitations. First, our analysis used provisional monthly data on suicide deaths, which may be corrected upward in subsequent months. However, the data used in this study were tabulated by the dates that the deceased was found, not by the dates that the deceased was presumed to have died, and thus the subsequent correction should be minimal. Second, our data on suicide deaths are aggregate in nature because NPA does not release individual-level data, thus our analysis could not control for the effects of confounders. Third, our online surveys relied on the commercial panel of respondents, who were not randomly sampled from the general population. Thus, the survey data might not include certain segments of the population, such as those with no access to the internet.
Despite these limitations, the present study makes an important contribution to the scientific community by reporting the trajectories of suicide deaths during the COVID-19 pandemic and by highlighting the most vulnerable populations during this unprecedented public health crisis. The experience in Japan may provide valuable implications for other countries. Given that the most affected industries were those mainly served by women in many other countries, it is likely that women constitute one of the highest risk groups for suicide in those countries as well. Similarly, since many countries also introduced school closures during the peak period of the pandemic, future studies should investigate the impact of school closures on schoolchildren and university students in Japan and in other countries. The results of our study strongly indicate that continuous monitoring of mental health and appropriate suicide prevention efforts are necessary during, and in the aftermath of, the COVID-19 pandemic.
Data Availability
The data on suicide deaths are public data and thus can be shared. The individual survey data cannot be shared due to the sensitive nature of the study.
Supplementary file1: Description of the online surveys
The status of mental health and economic conditions of the general population reported in this study were based on a series of ongoing monthly surveys. We sampled our respondents from a web panel of a major commercial survey company in Japan, the Survey Research Center. For each survey, the company sent out screening questions to approximately 10,000 registered individuals and then constructed a final sample of 1,000 respondents so that they were representative of the Japanese population in terms of their sex, age groups, and areas of residence.
Each of the surveys was conducted on the following dates:
1st survey – April 16-18
2nd survey – May 15-17
3rd survey – June 17-19
4th survey – July 17-21
5th survey – August 24-26
6th survey – September 23-25
Supplementary File 2: The results of regression-based analysis
In addition to the analysis presented in the main text, we have also run a simple linear regression model in order to check if the observed changes in suicide deaths in 2020 relative to the past years were statistically significant. We used monthly data between January 2017 and September 2020 for each demographic group, and thus the number of observations in each model was 45. The estimated model is as follows:
where the dependent variable was the number of suicides in month m in year y for group g, where m = 1, … 12. dk,2020 is an indicator variable for each month in 2020 (until September), where k = 1, …,9 and its associated coefficient, βk, captures the difference in the number of suicides in month k relative to the baseline period, which is the corresponding month in the last three years. The month fixed effects (i.e. twelve indicator variables for each month),γm, were included to capture the baseline monthly fluctuations in suicide deaths. The constant term was omitted from the model to avoid perfect collinearity and to make the interpretation easier. We report only the estimated coefficients on dk,2020 along with the 95% confidence intervals below, but the month fixed effects were always included in the estimation.
Note: the estimated coefficients on indicator variables for each month in 2020 along 95% confidence intervals shown. The data represent monthly data on suicide deaths in Japan between January 2017 and September 2020. The estimated models included month fixed effects.
Note: the estimated coefficients on indicator variables for each month in 2020 along 95% confidence intervals shown. The data represent monthly data on suicide deaths in Japan for both sexes between January 2017 and September 2020. The estimated models included month fixed effects.
Footnotes
Updated with the most recent data (September 2020).
