Suicide mortality rates in Japan before and beyond the COVID‐19 pandemic era

Abstract Statistical analyses from Japan reported increasing suicides in 2020, first in the world, proving the severity of the public health crisis during the COVID‐19 pandemic; however, so far, international suicides have not been shown to be objectively increasing at population level. Followed studies reported the existence of a substantial heterogeneity of suicides among subgroups and time‐lag impacts. Against public health crisis in Japan, policymakers, psychiatrists and public health personnel should prioritize improving suicide prevention programs following evidence‐based policymaking. Understanding how/what factors relate to the COVID‐19 pandemic and what other factors have shaped the increasing suicide numbers since 2020 through objectively well‐controlled/fine‐grained analyses of high‐quality longitudinal/cross‐sectional data at the individual, regional, and national levels is important for identifying the reasons for the recent trend. For this purpose, this study examined suicide statistics, statistical analysis methods, and their interpretations. Recent analyses suggest an increased suicide risk among females <50 years and males <30 years in 2020–2022. Notably, time‐series analyses revealed that adolescent suicides began increasing before the pandemic, while working‐age female suicides sharply increased synchronously with the pandemic outbreak. Causality analyses suggest that social issues facing Japan and recent global psychosocial and socioeconomic transformations are risk factors for suicide in high‐risk groups. Finally, this report demonstrates the importance of providing appropriate support based on an objective understanding of individuals who are at risk for suicide, without being bound by traditional established knowledges.


INTRODUCTION
Crude suicide mortality rates per 100,000 population (CSMRs) in Japan have varied widely over the past quarter-century.Until the early 1990s, the CSMR of Japanese males was ~17-18, which was not higher than other Organisation for Economic Cooperation and Development (OECD) countries. 1However, following the collapse of the asset bubble in 1991 and the 1997 Asian economic crisis, annual suicides in Japan increased drastically in 1998 (1997, 24,3921; 1998,   32,863).3][4] To address this crisis, the Japanese government enacted the Basic Act on Suicide Prevention in 2006 and the General

BASIS OF STATISTICAL MEASURES FOR SUICIDE MORTALITY TRENDS Importance of fluctuations in population/agedistribution
It is generally not incorrect to compare the annual/monthly suicides number in a region with the previous year, as long as the demographic composition (age-distribution/population) within the region does not change significantly.However, while suicide numbers in Japan decreased between 2020 and 2021 (21,081 to 21,007), 11 the CSMR increased from 16.58 to 16.59 due to the national population decreasing from 127,138,033 to 126,654,244. 33This discrepancy indicates the importance of population-based standardization of suicide mortality.
When the prevalence risk is independent of age, the regional age distribution is considered unimportant when comparing suicide statistics among regions.However, CSMRs disaggregated by age in Japan have increased in an age-dependent manner in the <60 years population. 7,8,12,28,34,35Therefore, when suicide risk is age-dependent and the age distribution in regions changes over time, age distribution becomes an important demographic variable for suicide statistics.In Japan, which has a decreasing birth rate and aging population (predominant in rural areas), even if CSMRs for each generation remain stable, national-level CSMRs may increase due to aging.For example, the Mie and Iwate prefectures are well known to have the lowest and highest CSMRs in Japan, respectively.In 2010, the female CSMRs in Mie and Iwate prefectures were the lowest (10.45) and the highest (22.17), respectively, whereas in 2018 they were 13.01 and 13.86, respectively (the second highest and highest CSMRs, respectively).
Conversely, the female SDRs of Mie and Iwate (based on population age distribution in 2009) were 12.78 and 12.18, respectively, with the female SDR of Mie prefecture being the highest in Japan, therefore suicide mortality standardized by age distribution is an important statistic.Most studies of international trends in suicide mortality provide standardized death rates using the WHO World Standard Population (WSP) model 36 or the European Standard Population (ESP) model. 37aditionally, the standardized mortality ratio in Japan has been calculated using the Japanese standard population model, which was based on the population from 1985 (1985JSP).However, the population of Japan is aging rapidly.Accordingly, in 2022, the model was revised based on the population distribution from 2015 (2015JSP). 38From now, SDRs will be calculated using the 2015JSP model.Therefore, the SDR in the present study was calculated using the 2015JSP model (Figure 1).

Analyses of temporal fluctuation in suicide mortality
2][43][44][45][46] To overcome these issues, recent studies have analyzed the temporal fluctuation or excess mortality of SDR/CSMR using four main models: (1) comparison between previous averages and observed data using an analysis of variance (ANOVA) or linear mixed-effect model (LMM), 23,28,47  (2)   comparison between predicted values calculated from the seasonal autoregressive integrated moving average (sARIMA) and observed value using ANOVA/LMM, 24,27 (3) detection of impacts of intervention (changing trends and discontinuities) using interrupted time-series analysis (ITSA), 13,20,25,47 and (4) detection of fluctuations of trends using joinpoint regression analysis (JPRA). 12,13,26,28[50] As a typical example, the Office for National Statistics in the UK publishes the average deaths of the previous 5 years as the average of past "normal" deaths. 49However, in the 5 years prior to the pandemic (2015-2019), Japanese suicide numbers continuously decreased by approximately 20% (from 24,025 in 2015 to 20,169 in 2019). 4,11It is necessary to carefully consider whether this 20% decrease is stable.In fact, when SDR/CSMR after the pandemic outbreak was compared with the average SDR/CSMR for 3 years (2017-2019) and 5 years (2015-2019) previously, the statistical analyzing results were quite different. 23,28Because the SDR/CSMR disaggregated by factors during the pandemic was found to be significantly higher than the average from 2015 to 2019, it can be concluded that suicides increased after the outbreak in that subgroup.However, the lack of a statistically significant increase does not mean that there was no increase in the risk of suicide.The increase during the pandemic may have been underestimated due to the strong decreasing trend in the prepandemic period. 28

Seasonal autoregressive integrated moving average model
The sARIMA is an established statistical method that is commonly used for national-level economic forecasting in various OECD countries, including Japan. 1,51We have already reported the comparison of suicides between predicted values calculated by the sARIMA and observed SDR/CSMR until 2021 using panel data. 23,24,27In particular, sARIMA exhibits strong advantages in future forecasting, including trends and seasonal variability. 52 female SDR/SMR, a transient decrease in the female SDR/CSMR at the pandemic outbreak, and a gentle upward shift in the SDR/CSMR of some groups of males and females. 23,24,27However, when there is markedly significant irregular variability in trends or seasonal variability in the observed period, the accuracy of the forecast is reduced. 52Considering that the pandemic lasted for 3 years, there are some issues to consider when implementing the sARIMA for the decreasing trends with large deviations, while the SDR predicted from 2009 to 2022 values shows increasing trends with relatively small deviations (Figure 2).

Time-series analyses
4][55] Based on these advantages, a number of studies have used ITSA to analyze the impacts of the pandemic on SDR/CSMR fluctuations. 13,20,25,47,56wever, ITSA is unable to detect any fluctuations that are unknown during the observation period.Thus, if a decreasing trend in suicide mortality has already changed before an intervention, ITSA may overestimate an increasing trend or positive discontinuity in fluctuations after the intervention. 13RA (also known as the segmented regression model) fits the simplest joinpoint model that the trend data allow and identifies significant points where trends change. 57JPRA is used to analyze the number of connection points by calculating the sum of residual error squares between the fitted value and the true value.The method uses model fitting to divide a long-term trend line into several trend sections. 12,26,28A comprehensive review of the statistics and underlying methodology applied in JPRA has been reported. 57The connecting points of different trend segments are called joinpoints.Thus, JPRA is a powerful statistical method for detecting unknown joinpoints (transformed trends and discontinuities). 57Compared with ITSA, JPRA can flexibly detect joinpoints in suicide mortality over observation periods but cannot directly detect the impacts of the target intervention.[41][42] The fluctuations in SDR from January-2009 to June-2023 detected using ITSA and JPRA indicated different temporal patterns (Figure 2).In males, ITSA detected decreasing trends in the male SDR until the pandemic (2009-2019), but there was positive discontinuity (sharply increasing) synchronized with the pandemic outbreak followed by an increase during the pandemic.However, JPRA detected two joinpoints in 2017 (decreasing to unchanging) and 2020 (unchanging to increasing) without positive discontinuity.
Interpreting the nonidentical temporal fluctuations of the male SDR detected by ITSA and JPRA, based on the statistical characteristics of ITSA and JPRA, male suicides showed a consistent decrease until 2016, which was attenuated from 2017 to 2019.ITSA probably overestimates the transformation from decreasing to increasing trends during the pandemic as a positive discontinuity due to the underestimation of attenuated decreasing trends before the pandemic (from 2017 to 2019).In females, ITSA also detected a consistent decrease in the female SDR before the pandemic and a larger positive discontinuity during the pandemic outbreak followed by no change during the pandemic.The effect size of positive discontinuity synchronized with the pandemic of the female SDR seemed to be larger than that of the male SDR.JPRA detected three joinpoints in 2011 (unchanging to decreasing), 2020 (decreasing to increasing), and 2021 (increasing to decreasing).ITSA may overestimate the sharply increasing trends after the pandemic as positive discontinuity (within 1 month increase during the pandemic), resulting in a gradual but significant decrease since 2021, which may be underestimated as unchanging.These findings indicate that JPRA may be suitable for analyzing the temporal fluctuations of SDR/ CSMR in Japan since 2009.

Exploring high-risk groups for suicide during the pandemic
The group at risk for suicide during the pandemic is almost the same as the group that showed an increased SDR/CSMR during the pandemic.When planning revised suicide prevention programs, it is important to clarify whether the increased SDRs/CSMRs of high-risk groups during the pandemic are due to pandemic-associated factors or other factors that occurred before or after the pandemic.Timeseries analyses can provide essential information for identifying highrisk groups by clarifying the bases for the temporal relationships between increasing SDRs/CSMRs disaggregated by different factors and the pandemic.

Basis of causality analyses for suicides
Recently, advances in computing power and software programs have made hierarchical linear models (HLMs, also known as multilevel models or split-plot designs) the most popular statistical method for analyzing suicide causality.HLMs are appropriate for designs where data for subgroups are organized in nested data.[60][61] HLMs can provide alternatives to univariate or multivariate analyses of repeated measures.Occasionally, regression analyses yield results that are difficult to interpret, called "Simpson's paradox," 62 where significant temporal trends are observed for each group (fixed effects) but disappear/reverse when all groups are combined (random effects). 63Random effects are unique, time-constant attributes of subgroups that are not correlated with the independent variables, whereas fixed effects are unique attributes of subgroups that do not vary over time.In particular, the HLM is considered to be a powerful application for determining fixed-and random-effects. 64ctor autoregressive analysis (VAR) is powerful statistical model that can be used to detect temporal relations between multiple quantities that fluctuate over time. 65VAR generalizes the singlevariable autoregressive model by allowing for multivariate time-series.
Similar to AR models, each variable is composed of an equation that model changes over time.This equation consists of the lagged values of the variable, the lagged values of the other variables in the model, and an error term.VAR is simpler than structural models with simultaneous equations, since it only requires lists of variables that can be hypothesized to affect each other over time. 6514]35 Granger causality is statistical hypothesis test for determining whether timeseries analysis is effective in forecasting the result. 66The impulse response explains the evolution of the model's variables in reaction to a shock in one or more variables. 67Forecast error variance decomposition -the extent to which the forecast error variance of each variable can be explained by exogenous shocks to the other variables in VAR-is used to aid in the interpretation of the VAR after fitting. 67Ms and VAR are both powerful causality analysis tools; however, in addition to suicide statistics, they require other large independent variables in large databases (e.g., government databases).Suicide statistics collected by the National Police Agency (SSNPA) and basic data on suicide in the region (BDSR) published by the MHLW are government suicide databases that publish monthly suicide numbers within the next month, which is fast relative to other countries. 11,68The SSNPA and BDSR are both considered to be as
In HLM analyses, before the pandemic, a random effect of CUR on CSMR could be detected in males in their 50s to 60s, but not in other groups of males or females. 8However, fixed-effects HLM analysis detected interesting but unexpected relationships between detected in both males and females. 28During the pandemic (2020-2021), positive fixed effects of CUR on male SDR but not female SDR were observed. 28Therefore, despite the similar temporal fluctuations between CUR and female SDR, the discrepancies in the fixed effects of CUR on female SDR before and during the pandemic suggest that the similarity in temporal fluctuation patterns can suggest the possibility of causality, but cannot fully confirm definitive causality.
Before the pandemic, CURs disaggregated by unemployment duration (0-3, 3-6, 6-12, 12-24, and >36 months) linearly decreased.All CURs disaggregated by unemployment duration, other than 36 months, showed positive discontinuity (sharply increased) synchronized with the pandemic, while CURs decreased during the pandemic, falling until they were almost equal to prepandemic values in 2022. 133][14] VAR with impulse response analysis detected a positive temporal relationship between 6-and 12-month periods of unemployment and the CSMRs of working-age males (20-69), the positive impacts persisted for over 2 years.In contrast, the positive temporal relationship between an unemployment period of <6 months and the CSMRs of females of 20-49 years persisted for approximately 1 year. 13These findings can provide the temporal causality underlying the increasing CSMRs of working-age females in the early pandemic period (from May to October 2020) and those of working-age males in the late pandemic period (2020).Therefore, even a relatively shorter period of unemployment had a positive impact on females suicide, which persisted for approximately 1 year.]28 In Japan, the duration for which unemployed individuals can receive unemployment benefits depends on the length of time that the individual has worked (i.e., individuals who have worked <1 year can receive benefits for 3 months and those who have worked >5 years can receive unemployment benefits for 1 year).Approximately 40% of part-time employees have worked for <1 year. 78The parttime employment rate in Japan is the highest (39.1%) among OECD countries (OECD average 25.3%). 79Part-time employees are predominantly female in Japan, 78  women's labor force participation, including low-skilled/low-wage women (nonregular workers), is considered to be the underlying mechanism of the "gender paradox." 73,74Low-skilled/low-wage workers are usually the most expendable during an economic recession; consequently, women may be as affected as men (or more so) by deteriorating employment conditions. 73,74e discrepancy between the Asian Financial Crisis (1997-1998)   and the COVID-19 pandemic (2020-2023) might be related to the transformation of the labor force participation structure in Japan. 13en labor force participation rates of females are low, the impacts of increasing CUR on females SDR/CSMR cannot be observed, whereas increasing females labor force participation has been reported to contribute to the positive association between CURs and CSMRs, resembling that of males. 80In Japan, the labor force participation rates of females during the Asian Financial Crisis (1997-1998) were approximately 60%, but they began increasing in the late 2010s and were >75% during the pandemic (those of males remained ~85% over time). 51,81Taken together, the recent increase in female labor force participation may play an important role in the discrepancy between insensitivity of female CSMRs to increasing CURs during the Asian economic crisis and the converse high sensitivity observed during the early stage of the pandemic. 13

Temporal fluctuation of adolescent CSMRs caused by suicidal motives
Suicide is the fourth leading cause of death among 15-29-year-olds (adolescents) worldwide and is even more serious among youths in Japan, where it remained the leading cause of death among youths between 2009 and 2021. 31,82Adolescence is a period of psychosocial and biological development that involves various social statuses (e.g., middle school, high school, university, and special vocational school students and working young adults). 835][86][87][88] In contrast, in Japan, CSMRs of males and females of 10-19 years began to increase before the pandemic. 12,25,26,28To date, there have been no causality analyses using independent variables that may affect the specific temporal fluctuations of adolescent CSMRs.However, temporal fluctuations in CSMR caused by suicidal motives may suggest, at least partially, factors underlying the increasing adolescent CSMRs.
It has been reported that schools, at least partially, contribute to suppressing adolescent CSMRs in Japan. 25CSMRs disaggregated by school age increased with age from middle school, high school, and university to special vocational school.The CSMRs of special vocational school students were almost equal to those of workingage generations. 25,26Notably, the CSMRs of male and female middle school students were almost equal, but the age-dependent increase in CSMRs among male students was more pronounced than that among female students. 25,26The primary causes of student CSMRs were worrying about the future and underachievement (in schoolrelated problems) and mental illness (depression and other mental illnesses). 25,26School-related problems were more impactable for CSMRs of male students than mental illness, whereas mental illness was more impactable for female students.Indeed, CSMRs caused by depression and other mental illnesses in middle-school and highschool females were greater than those in males, whereas the CSMRs caused by mental illness in males and females were almost equal in university and special vocational school students. 25,26Considering that the onset of internalizing disorders occurs at approximately 15 years of age and the higher prevalence in female adolescents in comparison to males, internalizing symptoms/disorders probably play important roles in the greater CSMRs caused by these disorders in female students. 25,26,89,9026,28 The GPSPP establishes suicide prevention guidelines that are promoted by the government and have been revised approximately every 5 years (first: 2007-2012; second: 2012-2017; third: 2017-2022) according to the Plan-Do-Check-Act cycle. 91To respond to adolescent suicide, the GPSPP has been revised to add the following priorities: "development of mental health support service in schools" in the first GPSPP (2007-2012) term, "enhancement of support and counseling systems for bullied children and victims of child abuse or sex crimes" in the second GPSPP (2012-2017), and "education on how high-risk youth can request support" and "development of a suicide prevention program for children/adolescents" in the third GPSPP (2017-2022). 25,91Therefore, the suicidal motives of adolescents, different from those of other generations of working-age and the elderly, were insensitive to GPSPP. 25,26,92e impacts of the three GPSPP periods on student CSMRs disaggregated by suicidal motive from 2007 to 2022 was analyzed by ITSA with a dual interventions model (2012 and 2017). 25The CSMRs of high school and university students remained unchanged during the first GPSPP and decreased during the second GPSPP.However, they drastically increased during the third GPSPP.The analysis revealed similar fluctuation patterns of CSMRs caused by schoolrelated (underachievement) and health-related (depression and other mental illness) motives in high school and university students of both sexes during the third GPSPP.CSMRs caused by conflict with parents consistently increased throughout the observation period in female high school and male university students.ITSA with a triple interventions model that added the pandemic in 2020 to the GPSPP periods detected additive impacts of the pandemic on increasing trends of some CSMRs during the third GPSPP, whereas it showed no independent impact on the increasing trends of student CSMRs during the third GPSPP.
The third GPSPP (2017-2022) listed new priority categories: "development of suicide prevention program for children/adolescents," involving the enhancement of support systems for bullied/ abused student/children in the school/community and "enlightenment of how to represent signs for seeking support against highrisk student/children." 91The 2022 White Paper on Suicide Prevention published by the MHLW reported that >25% of adolescents who suffered from depression or suicidal ideation were unable to seek nearby support by themselves. 31The third GPSPP captured future problems accurately.Although efforts to suppress CSMRs in adolescents/students have been made, the fact that these rates have not decreased suggests that the recent deterioration in psychosocial development or increasing internalizing disorder/symptoms in adolescents/students may play more significant roles in adolescent/student suicides than initially expected. 25,26In addition to new additional priorities in the third GPSPP, enlightenment on how signs of seeking support are represented in high-risk student/ children, and enlightenment of schools, families and communities to the fact that students/children who are at risk due to internalizing symptoms/disorders cannot represent signs may also be important.

OTHER POSSIBLE FACTORS Werther's effects and anomie
In the second half of 2020, female suicides by hanging at home increased regardless of region. 23,24The MHLW repeatedly reminded various media sources to report on celebrity suicides in compliance with WHO guidelines 93 (18 times during the pandemic). 94The MHWL estimated that increasing hanging suicides are probably imitative suicides. 95Increasing hanging suicides were observed from 1 month later and persisted for several months after celebrity suicides, indicating the characteristics of copycat suicides triggered by frequent reports of suicide of celebrities from mass media. 23,24,95though the annual number of female suicides by hanging increased in 2020 (n = 47) compared with the average for 2015-2019 (n = 30.8± 2.4), 11 the rise in suicides by hanging alone could not completely account for the increase in female suicides in 2020. 23,24teraction between imitative suicide (Werther's effect in a narrow sense) 96 and anomie suicide (norm-disruptive events) rooted in Durkheim's theory 97 has been proposed as the main explanatory mechanism for Werther's effect, 95,98 and drawing conclusions about these complex relations remains a challenge. 99Durkheim hypothesized that in both economic recessions and booms, social systems are unable to sufficiently adapt to individual needs, leading to increasing suicides via weakened social integration. 80,97,983][14] A study reported that the outbreak and end of the pandemic, including social restrictions, had heterogeneous reported that the delay in digitalization in Japan was a serious socioeconomic issue before the pandemic.However, during the pandemic, digitalization rapidly progressed out of necessity. 101Indeed, in 2022, the CSMRs of males (caused by failure at work) and females (caused by trouble adjusting to changing work environments) drastically increased. 4,13,14,31,68Although it is easy to understand that the aging population in Japan contributes to the lower degree of adaptation to changes in social structure (including rapid digitalization), further analyses should be conducted to investigate CSMRs caused by employment-related motives since 2023.

Decreasing birth rate with aging
The long-lasting decrease in birth rates with aging is serious social problem in Japan that has led to the aging of the demographic structure with decreasing labor force, as well as an increasing number of only-children. 101,1020][61] It can be speculated that governmental expenditure shifted to public health during the pandemic.The publication of reports on national and regional governmental expenditure from 2020 to 2023 is awaited to clarify the relationships in detail.
Although it is outside the scope of this review, a brief discussion regarding the decreasing birth rates as remaining challenges for the future should be included.Statistical analyses indicated that internalizing symptoms/disorders play important roles in increasing adolescent/student suicides. 25,26Long-lasting, progressively decreasing birth rates have continuously increased for only-children without siblings, with rates of >20% in 1997 and >30% in 2015. 102,112storically, the emergence of second children/siblings had been argued to contribute to developmental crisis for firstborn children by earlier psychodynamics 113,114 ; however, while the birth of a second child/sibling probably provides challenges to very young firstborn children it may also lead to rapid development.0][121] Therefore, children raised with siblings may be less affected by the stressful environment and/or gain flexibility. 119A recent Chinese study reported that scores associated with internalizing, externalizing, emotional, and behavioral problems of only-children were higher than those of firstborn children with siblings. 120 is easy to understand that parents tend to adopt strategies of high-care/high-control in relation to an only-child. 25,26,113,114Parker categorized parenting attitudes into two dimensions, "care" (consideration, concern, and affection expressed by the parent) and "control" (intrusiveness, protection, and manipulation unrelated to affectionate feelings). 1223][124][125][126][127][128][129][130][131] The depressive mood and hopelessness of university students are positively related to "high-control" and "low-care." 132Furthermore, mothers with internalizing symptoms/ disorders tend to have attitudes with affectionless control styles (lowcare/high-control). 132 These findings suggest the possibility that maternal internalizing symptoms/disorders may induce a vicious negative cycle that increases the prevalence of internalizing disorders in later generations. 25Indeed, the patient survey published by the MHLW reported an increasing prevalence of internalizing disorders in individuals of 10-24 years in 2020 compared with 2017. 25,26,133 Indeed, the combination analysis of the sARIMA with multivariate analysis of variance/LMM detected turbulence of seasonal fluctuations in the F I G U R E 1 Comparison of age distribution among international standard population models and age-standardized suicide death rates per 100,000 in Japan from January 2009 to June 2023 calculated using international standard population models, respectively.1985JSP, Japanese standard population model based on the population in 1985; 2015JSP, Japanese standard population model based on the population in 2015; CSMR, crude standardized suicide mortality rate per 100,000; ESP, European standard population model; SDRs, suicide death rates; WSP, WHO world standard population model.SUICIDE MORTALITY RATES IN JAPAN | 3 of 14 predicted values in the target period.Due to the statistical characteristics of the ARIMA (predicting values based on previously observed data), the longer the prediction period is, the greater the error in the predicted values.When the SDR for 2024-2025 was predicted using the sARIMA based on values from 2009 to 2019 and 2009 to 2022, the results were quite different (Figure 2).The predicted SDR for 2024-2025 based on 2009-2019 values shows F I G U R E 2 Fluctuations of annualized monthly SDRs in Japan.Predicted SDRs calculated by sARIMA using observed monthly SDRs (a) from 2009 to 2022 and (b) from 2009 to 2019.The red and blue regions on the right of each graph indicate the 95% confidence intervals.Fluctuations of the SDRs of (c) males and (d) females calculated by ITSA and JPRA from January 2009 to June 2023.Blue and red lines indicate the results calculated by JPRA and ITSA, respectively.Solid and dotted lines indicate the significant and not significant trends of SDR detected by statistical analyses (JPRA and ITSA), respectively.Green lines indicate the adjustment for seasonality by ITSA.In all panels, gray circles indicate the observed annualized monthly SDR values.ITSA, interrupted time-series analysis; JPRA, joinpoint regression analysis; sARIMA, seasonal autoregressive integrated moving average; SDR, suicide death rate.
of 30-69 years of age were attenuated prior to the pandemic; (2) CSMR trends of males <20 years began to increase prior to the pandemic and continued increasing throughout the pandemic; (3) CSMR trends of females <30 years began increasing prior to the pandemic and sharply increased synchronously with the pandemic, followed by unchanging trends during the pandemic; (4) CSMR trends of males of 40-69 years of age increased throughout the pandemic; (5) decreasing CSMR trends of males of 70-79 years and females in their 70s and 80s were attenuated during the pandemic; (6) CSMRs of males of 20-29 years and females of 20-49 years sharply increased synchronously with the pandemic, followed by a decrease during the pandemic; and (7) CSMRs of males of 60-69 and >80 years increased in the late phase of the pandemic (Figure 3).These findings suggest the existence of three types of suicidal vulnerability after the late 2010s.CSMRs of people under 20 years old began increasing before the pandemic, suggesting that risk factors unrelated to the pandemic were important for increasing suicides.CSMRs of males of 20-29 years and females <70 years sharply increased early in the pandemic but did not increase in the late stage.The CSMRs of males of 40-69 years probably increased during the late stage of the pandemic, therefore pandemic-associated risk factors for suicide may include factors related to the pandemic and the end of the pandemic.

F I G U R E 3
Fluctuations of annualized monthly CSMRs from January 2009 to June 2023 disaggregated by sex/age in Japan: (a) <20 years, (b) 20-29 years, (c) 30-39 years, (d) 40-49 years, (e) 50-59 years, (f) 60-69 years, (g) 70-79 years, and (h) >80 years.Blue and red circles annualized monthly CSMRs of males and females, respectively.Blue and red lines indicate the results calculated males and females CSMRs by JPRA, respectively.Solid and dotted lines indicate the significant and not significant trends of CSMRs detected by JPRA, respectively.CSMR, crude suicide mortality rate; JPRA, joinpoint regression analysis.
accurate and reliable suicide statistics databases.In Japan, judicial police must investigate the personal characteristics and background factors of each suicide case.The police investigate suicide motives based on evidence, suicide notes, official documentation (e.g., medical certificates and clinical recordings), and testimony from the victim's family.The results of their investigation discuss the different motives for suicide, which are compared with previously compiled lists of motives for suicide (52 subcategories in the SSNPA). 11,14,22,25,26,68Time-series analyses of temporal fluctuations of CSMRs disaggregated by suicidal motives in the SSNPA or BDSR can demonstrate temporal variation in causality, albeit indirectly through temporal fluctuation in suicidal motives.
CUR and SDR/CSMR before and during the pandemic.Before the pandemic (2009-2019), positive fixed effects of CUR on SDRs were F I G U R E 4 Fluctuations of annualized quarter SDRs and CURs of (a) males and (b) females in Japan from January 2009 to June 2023.Blue and red circles indicate the observed annualized quarter SDRs of males and females, respectively.Blue and red lines indicate the results calculated males and females SDRs by JPRA, respectively.Gray circles indicate the quarter CURs of males and females.Black lines indicate the results calculated males and females CURs by JPRA.CUR, completely unemployment rate; JPRA, joinpoint regression analysis; SDR, suicide death rate.
therefore part-time female employees suffer from shorter cycles of employment and have shorter periods of unemployment benefit.The relationship between employment structures and unemployment benefit periods in Japan can plausibly indicate temporal causality from the increasing number of unemployed females who had worked for a shorter period of time in relation to the rapidly increasing CSMRs of females in from 20 to 49 years from the third to fourth quarters of 2020.A similar phenomenon of greater sensitivity of female suicides to economic recession in comparison to male suicides was observed in Hong Kong in 1997-2003 and is described as a "gender-paradox." 74Increasing SUICIDE MORTALITY RATES IN JAPAN | 9 of 14 psychological effects. 100Early in the pandemic, individuals suffered from stress due to forced drastic changes in lifestyle and social systems.Meanwhile, since late 2022, individuals have been forced to prepare to adapt to the "new normal" in the post-COVID-19 period due to the prospect of ending the pandemic.The 2022 White Paper Information and Communications in Japan published by the Ministry of Internal Affairs and Communications

CONCLUSION
Based on suicide statistics in Japan from January 2009 to June 2023, this review discusses the mechanisms underlying decreasing suicides in 2009-2019 and increasing suicides in 2020-2023.Political authorities in Japan rapidly implemented enhancements of unemployment measures, support for small-to-medium-sized enterprises, and mental/social support programs to respond to psychosocial/socioeconomic deterioration by leveraging existing comprehensive suicide prevention programs, according to concerns of economics, psychiatry, and public health.Despite these efforts, suicides increased for 3 years during the pandemic.This review suggests that traditional established suicide risks alone cannot fully explain the increasing number of suicides in Japan since 2020.In particular, the recent increase in the social participation rate of females has played an important role in the increasing number of suicides among working-age females via the enhanced sensitivity of female suicides to short-term unemployment during the initial stage of the pandemic.In contrast, male suicides, which were relatively stable during the initial stage of the pandemic, increased in the second half of 2022.It cannot be denied that in addition to longterm unemployment, the drastic changes in the workplace during the pandemic, including digitalization, may have contributed to this increase.Most importantly, increasing internalizing symptoms/ disorders in adolescents/students due to decreasing birth rates/increasing only-children or parenting attitudes (high control) has played an important role in increasing adolescent/student suicides since the late 2010s.