Association of socioeconomic factors with hikikomori in Japan from 2010–2019: a correlational study


 Background

Although several surveys have suggested that socioeconomic factors function as background variables for the expression of hikikomori (prolonged social withdrawal), no research has directly examined whether socioeconomic indicators increase the occurrence of hikikomori. This study aimed to examine the relationship between socioeconomic factors and the expression of hikikomori using data covering a 10-year period.
Methods

We collected socioeconomic data between 2010 to 2019 from members of a multi-branch Japanese association for family members of hikikomori patients.
Results

The results of the partial correlation analysis and meta-analysis showed that the higher the unemployment rate, the greater the occurrence probability of hikikomori.
Conclusion

The findings suggest that socioeconomic factors relate to the increase in hikikomori, and that these factors should be included when identifying the individual or cultural factors that cause hikikomori.


Abstract Background
Although several surveys have suggested that socioeconomic factors function as background variables for the expression of hikikomori (prolonged social withdrawal), no research has directly examined whether socioeconomic indicators increase the occurrence of hikikomori. This study aimed to examine the relationship between socioeconomic factors and the expression of hikikomori using data covering a 10-year period.

Methods
We collected socioeconomic data between 2010 to 2019 from members of a multi-branch Japanese association for family members of hikikomori patients.

Results
The results of the partial correlation analysis and meta-analysis showed that the higher the unemployment rate, the greater the occurrence probability of hikikomori.

Conclusion
The ndings suggest that socioeconomic factors relate to the increase in hikikomori, and that these factors should be included when identifying the individual or cultural factors that cause hikikomori.

Background
Socioeconomic factors such as unemployment are commonly related to mental health problems. For example, previous studies have reported that the number of suicides is related to the unemployment rate in various countries [1]. Hikikomori or prolonged social withdrawal is also a mental health problem and can be presumed to be related to socioeconomic factors. Hikikomori -a phenomenon characterized by avoidance of social interactions for more than six months [2]-has been recognized as a serious social problem in Japan for over 20 years [3,4]. Hikikomori has a signi cant negative impact on psychological and physical health, such as comorbid psychiatric and nutritional disorders, leading to a deterioration of an individual's quality of life [5][6][7][8].
Hikikomori or similar phenomena have been reported in various other countries, including France [9], Australia, Thailand, Taiwan, Bangladesh, Iran, India, Korea, the USA [10,11], Oman [12], and Spain [13]. An epidemiological study of hikikomori in Japan in a community-based population aged 20-49 years revealed that 1.2% of that population had experienced the phenomenon in their lifetime [14], while a cross-sectional telephone-based survey in Hong Kong revealed that 1.9% of 12-29-year-olds had experienced it [15].
It can be assumed that biological, psychological, and social factors relate to the expression of hikikomori, although its causes are not yet well understood. In terms of biological factors, hikikomori can cause other psychiatric disorders and vice versa [16]. Kondo et al. [5] found that 80.3% of individuals with hikikomori were diagnosed with psychiatric disorders such as developmental, mood, anxiety, and personality disorders. Furthermore, Hayakawa et al. [17] found that hikikomori was related to serum uric acid levels in men and high-density lipoprotein cholesterol levels in women. Regarding psychological factors, previous studies have found hikikomori to be related to a loss of secure attachment during childhood [18], di culties in interpersonal relationships [19], depression and anxiety [20], Internet addiction [21], and loneliness [11]. Among social factors, many researchers have examined the relationship between hikikomori and familyrelated factors, because individuals with hikikomori spend most of their time at home. However, while some studies have found an association between family dysfunction and hikikomori [20,22], others have reported that the parents' childrearing style was unrelated to the expression of hikikomori [23], and that family interactions in hikikomori and non-hikikomori cases did not differ [24]. These ndings suggest that this issue still requires further research and investigation.
Socioeconomic factors may be considered as major external factors related to hikikomori. For example, the results of a community-based survey of Japanese residents aged 15-39 years conducted by the Japanese Cabinet O ce reported that many cases of hikikomori have work-related causes, such as poor work environment and being currently unemployed [25] ; another Cabinet O ce study of residents aged 40-64 years found that the most common reason for social withdrawal was leaving one's job (17 out of 47 hikikomori patients [26]. This implies that other socioeconomic factors, such as the unemployment rate [27], number of active job openings, and average household income, relate to the expression of hikikomori. However, previous reports have only examined individual reasons for hikikomori, and it is not clear whether broader socioeconomic factors in uence the increase in hikikomori cases. It can be presumed that employees often leave a company due to worsening socioeconomic factors, poor physical and mental health, and/or poor relationships with others in the workplace. To the best of our knowledge, however, no research has been conducted to examine the relationship between socioeconomic indicators and the increase in the occurrence of hikikomori.
We investigated over 20 years of data from a Japanese association of family members of hikikomori patients to examine the relationship of socioeconomic indicators, such as the unemployment rate and active job openings ratio, with the expression of hikikomori. Furthermore, for comparison with hikikomori, we investigated the relationship between socioeconomic factors and the number of occurrences of students not attending school-a concept similar to hikikomori, characterized by students who do not or cannot attend school for 30 days or more per year, mainly due to psychological, emotional, physical, or social factors, excluding cases of illness or those involving nancial reasons [28].

Data collection
We used data covering the years 2010 to 2019 from members of the KHJ (Kazoku [meaning "family" in Japanese] Hikikomori Japan) or the National Federation of Families with Hikikomori Persons in Japan, an association with more than 50 branches throughout the country. While the data collected from KHJ members included those in the year 1999 onwards, some items used in the current study only included those beginning in 2010. Some of the data used in the present study were also used by Nonaka et al. [29], Makabe et al. [30], Nonaka et al. [31], Nonaka et al. [32], and Nonaka, Shimada, and Sakai [24], to investigate family characteristics in cases of hikikomori. We used the age of the individuals with hikikomori and the age at the onset of hikikomori to calculate the number of expressions of hikikomori each year. Furthermore, to be eligible to participate in this study, participants were required to provide their age and gender, as well as those of the family member with hikikomori and the duration of the syndrome. If hikikomori occurred more than once, participants provided the duration and age at onset for each instance.

Procedure
Participation in the family association involved monthly meetings, during which family members were asked to participate in the investigation; questionnaires were then sent by postal mail. Informed consent was obtained before the participants responded to the questionnaire. The participants answered the questionnaires anonymously and were free to withdraw from the study at any time. The study was approved by the local research ethics committee of the institutes to which the authors belong and followed the principles of the Declaration of Helsinki which excluded research pre-registration in a database.

Instruments
Demographics. Participants reported their age; relationship with the individual with hikikomori (father, mother, or other, including sibling); and the age, gender, and duration of the syndrome for the individual with hikikomori [2]. We calculated the year at which the individuals' hikikomori rst occurred by subtracting the age of the individuals at the onset of hikikomori from their age at the time of data collection". For example, in the case of research conducted in 2018, if a 30-year-old individual rst experienced hikikomori at age 20, we determined that hikikomori was initially "expressed" in 2008 by using the following formula: 2018 -(30 − 20) = 2008. Then, we counted the number of expressions of hikikomori every year to examine its relationship with the socioeconomic situation.
Socioeconomic indicators. We used Japan's unemployment rate [27], active job opening rate [33], and average household income [34] for each year of the study as socioeconomic indicators. The unemployment rate is the proportion of unemployed persons (persons who have no jobs, are looking for jobs) in the labor force (people aged 15 and over who can currently work). The active job opening rate is calculated by dividing monthly active job openings by monthly active job applications.
Number of instances of school non-attendance by children at elementary and junior high schools. We used the number of children not attending Japan's elementary and junior high schools [35] as a phenomenon analogous to hikikomori because it is characterized by the avoidance of social interaction, and examined its association with socioeconomic factors. These students were de ned as those who do not or cannot attend school for 30 days or more per year, mainly due to psychological, emotional, physical, or social factors, excluding cases of illness or those involving nancial reasons [28]. We calculated the number of annual occurrences of school-non-attending students by excluding the number that had continued from the previous year from the current number of instances. As school non-attendance has various evidenced background factors such as mental health and family member in uence [28], we presumed that it was not strongly related to socioeconomic factors.

Data Analysis
We performed correlation analyses to examine the association of socioeconomic indicators with the expression of hikikomori, unemployment rate, active job opening rate, average household income, and number of expressions of hikikomori per year. We excluded outliers in the year of hikikomori onset for each person. We performed a random effects meta-analysis using the restricted maximum-likelihood estimator (REML).

Results
Participant demographics Table 1 shows the age and gender of the family respondents, the age and gender of the persons with hikikomori, and the duration of hikikomori for each year the research had been conducted. There were 2,461 family respondents across the 8 instances of data collection over 10 years. The average age of individuals with hikikomori across the period of data collection was 30.94 to 34.78, and the duration of hikikomori was 112.80 to 144.32 months.

Correlation between socioeconomic indicators and expression of hikikomori
The plots of the number of expressions of hikikomori correlated with the unemployment rate ( Fig. 1), active job opening rate (Fig. 2), and average household income (Fig. 3) are shown by research year. Table 2 shows the correlation analysis results for the socioeconomic indicators and the expression of hikikomori. Spearman's ρ was .  Table 3). The unemployment rate was signi cantly correlated with the expression of hikikomori except for the research in 2018. Note. ρ: Spearman's rho. The results of the meta-analyses indicated that the partial correlation coe cient was .44 (95% CI .34 to .55, p < .001; Q (7) = 5.92, p = .55; I 2 = 0.00%) for the unemployment rate (Fig. 4); − .09 (95% CI − .22 to .04, p = .17; Q (7) = 6.07, p = .53; I 2 = 0.00%) for the active job opening rate (Fig. 5); and .24 (95% CI − .01 to .49, p = .06; Q (7) = 43.35, p < .001; I 2 = 82.22%) for the average household income (Fig. 6). The effect sizes of the unemployment rate were medium, while the effect sizes of the average household income and active job opening rate were small [36]. The household income showed high heterogeneity. These results suggest that unemployment increases the likelihood of hikikomori occurrence, but that an increase in job openings alone may not be enough to prevent hikikomori.

Correlation between socioeconomic indicators and school non-attendance
Children's school non-attendance showed a signi cant partial correlation coe cient with unemployment rate (ρ = .49, 95% CI .09 to .75, p = .02), but not with active job opening rate (ρ = .24, 95% CI − .19 to .59, p = .28) or household income (ρ = − .24, 95% CI − .59 to .19, p = .28). Spearman's correlation coe cient between expression of hikikomori data from 2010 to 2019 and school non-attendance was − .11 to .52, and the results of the meta-analysis were .19 (95% CI .01 to .37, p = .04; Q (7) = 10.58, p = .16; I 2 = 37.47%). The correlation coe cient for the unemployment rate was the same as that for the expression of hikikomori and was not signi cant for the active job opening rate and household income.

Discussion
This study aimed to examine the relationship between socioeconomic factors and the expression of hikikomori. The data were collected 8 times over a period of 10 years, and have a major limitation in that it is not clear how many participants overlapped between the data collection points (i.e., annually). Even with this limitation, the ndings of the current study are important, as it is the rst study to demonstrate a relationship between socioeconomic factors and the expression of hikikomori, which affects a reclusive and hard-to-access clinical population.
In the current study, many individuals with hikikomori were long-term cases, compared with previous research targeting 18-to 34-year-olds in France (mean duration = 29 months; [9]) and 16-to 35-year-olds in Japan (mean duration = 4.3 years; [5]). However, recent epidemiological studies in Japan have shown that 34.7% of individuals with hikikomori aged between 15 and 39 years have had the disease for more than 7 years [25,37], while 38.2% of those aged between 40 and 64 years have had hikikomori for more than 7 years [26]. The duration of hikikomori in this study was similar to the aforementioned ndings. The gender ratio of about 80% male was also similar to that in several previous studies in Japan and elsewhere-a ratio of 76.6% male was determined by the Japanese Cabinet O ce [26], 80% by Chauliac et al. [9], 76.4% by Funakoshi and Miyamoto [38], 74.8% by Kondo et al. [5], and 75.6% by Lee et al. [20]. A possible reason for this gender ratio is the Japanese socio-cultural background, in which men typically experience more workrelated psychological pressure than women do [39]. School non-attendance showed almost the same correlation coe cient as hikikomori. The relationship between school non-attendance and unemployment rate supports previous research [40]. As both showed a low correlation coe cient, further research is needed to clarify the reason for this. For example, school non-attendance, unlike hikikomori, may occur indirectly due to economic and psychological changes in the family as a result of parental unemployment, and may often involve antisocial behavior rather than asocial behavior.
Previous studies have shown that unemployment was the most common cause of hikikomori for residents aged 40-64 years [26]. The current results suggest that a preventive approach to long-term hikikomori could involve mental health and reemployment support for the unemployed population. On the other hand, the active job opening rate, which is the number of jobs available divided by the number of jobseekers, an important indicator of overall employment conditions, did not have a high correlation with hikikomori. As individuals with hikikomori often stop seeking jobs [26], the active job opening rate may not be related to the number of hikikomori cases. Considering that hikikomori cases may not necessarily occur in all countries with a high unemployment rate, sociocultural factors such as family relationships and psychological factors such as personality traits may serve as moderators of the expression of this effect.
Average household income was also not linked to the expression of hikikomori based on the meta-analysis results. This stands to reason, as it is di cult to maintain hikikomori without the support of a household income. However, the results showed that the correlation coe cient of household income varied widely from year to year, indicating no clear correlation between household income and the expression of hikikomori. In addition, considering that Yong and Nomura [19] reported that social class was not signi cantly different between hikikomori and non-hikikomori cases, further research is needed to clarify the relationship between the family economic situation and expression of hikikomori. The data representing average household income exhibited heterogeneity, and the partial correlation coe cient was different for the data collected There are several limitations to this study. First, while the results showed the relationship between socioeconomic factors and expression of hikikomori, we could not determine causation. Future research is also needed to determine whether other variables, such as mediators and moderators, further affect this relationship.
Second, sample bias is an important limitation. As most of the participants were concerned with solving the problems of their hikikomori family members, the possibility of sample bias cannot be excluded. It is important to note, however, that since individuals with hikikomori are likely to avoid interpersonal situations including treatment [42], research targeting family members may be more expedient and may potentially reduce the effect of the bias of the person with hikikomori. Telephone-based surveys [15] and online counseling via e-mail [43] may also be useful in reducing bias, due to the hikikomori characteristic of avoiding going out, although it may be di cult to reduce the bias due to avoidance of other people.
Clinically, family training aimed at easing access to individuals with hikikomori has shown its e cacy [44][45][46], and may be helpful for engaging the individual with hikikomori in treatment.
Issues with research design robustness are also a limitation of this study. Although long-term follow-up is desirable, dropouts are likely to occur in large-scale follow-up studies, especially given the reclusiveness of hikikomori, or in cases when hikikomori is resolved. This may help explain the dearth of longitudinal studies of hikikomori, though one 12-month follow-up study has been reported in recent years [42]. One way to solve this problem is through Internet-based research; however, considering that the proportion of individuals with hikikomori (29.8%) who use the Internet on a daily basis is lower than that of non-hikikomori individuals (43.3%) [26], this may introduce another source of bias. Moreover, it must be considered whether the ndings of this study are applicable outside of Japan, especially in less economically developed countries.
Naturally, in addition to socioeconomic factors, various other biological, psychological, and social factors will in uence the expression of hikikomori. Further comprehensive studies are needed to clarify the in uence Page 10/19 of these additional factors.

Declarations
Ethics approval and consent to participate The study was approved by the local research ethics committee of the institutes to which the authors belong.

Consent for publication
Not applicable Availability of data and materials Not applicable

Competing interests
The authors have no con icts of interest directly relevant to the content of this article.

Funding
This work was supported by JSPS KAKENHI Grant Numbers 20730447 and 18H05819, and by the Ministry of Health, Labour and Welfare of Japan. The funding source had no role in the study design; the collection, analysis, and interpretation of data; the writing of the report; or the decision to submit the article for publication.
Authors' contributions S.N. conducted the analysis and wrote the manuscript. Both authors contributed to study design and data interpretation, and reviewed the manuscript.  Forest plot of the combined correlation of the unemployment rate and the expression of hikikomori.

Figure 5
Forest plot of the combined correlation of the active job opening rate and the expression of hikikomori.

Figure 6
Forest plot of the combined correlation of the average household income and the expression of hikikomori.