Introduction
Hikikomori is a word deriving from the Japanese “hiku” (to pull) and “komoru” (to withdraw),Reference Martinotti, Vannini and Di Natale1 that in the last decades has become widely used for indicating a long-lasting social withdrawal that cannot be better explained by other psychiatric disorders.Reference Kato, Kanba and Teo2, Reference Kato, Kanba and Teo3 Initially, hikikomori was considered a phenomenon proper of Japanese culture, linked to specific sociocultural stress factors. It was thought to be a way of escaping, used by younger people, in relation to the growing social pressure and to the feeling of being incapable to fulfill the expected life goals.Reference Furlong4 Recently, this condition has been recognized and documented in different areas of the world, and the literature has grown with the description of cases coming from many countries, especially in high-income ones.Reference Harding5-Reference Wu, Ooi, Wong, Catmur and Lau12 Hikikomori is commonly described as a form of pathological social withdrawal or social isolation that lasts for more than 6 months and causes significant impairment in functioning and/or subjective distress.13-Reference Kato, Kanba and Teo15 Moreover, the symptomatology should not be caused or influenced by an underlying psychosis, while residual social contact through the Internet is contemplated.Reference Kato, Kanba and Teo14, Reference Kato, Kanba and Teo15 Due to the core characteristics of this condition, carrying an accurate epidemiological study can be difficult; indeed, by definition, these patients experience a severe social withdrawal which often prevents clinical contact. However, a recent study carried out on the European population estimated a hikikomori prevalence of 1.71%.Reference Amendola, Cerutti and von Wyl16 Another issue in the timely recognition of this syndrome is due to the high rates of comorbidity. Indeed, in over half of the cases, hikikomori is found to be comorbid with various psychiatric disordersReference Koyama, Miyake and Kawakami17including various anxiety disorders, major depression, personality disorders, Internet Gaming Disorder, and Autism spectrum disorder (ASD).Reference Lee, Lee, Choi and Choi18-Reference Teo, Stufflebam and Saha21 While the literature about the link between hikikomori and ASD is still limited, a recent study suggested that hikikomori manifests in up to one-third of ASD cases and autistic-like features are often described in hikikomori subjects.Reference Tateno, Park, Kato, Umene-Nakano and Saito20, Reference Dell’Osso, Dalle Luche and Maj22-Reference Teo and Gaw25
As defined by the Diagnostic and Statistical Manual for Mental Disorders (DSM), ASD is a neurodevelopmental disorder defined as a pervasive pattern of deficits in verbal and nonverbal communication and social interactions, associated with restricted interests, repetitive and stereotyped behaviors, and impaired sensoriality.13 While the more severe forms of ASD are often easily detectable from early childhood, the milder forms, without intellectual or linguistic disability, can often remain unrecognized for many years and come to clinical attention only in adulthood, after the development of other comorbid psychiatric symptoms, often leading to misdiagnosis.Reference Dell’Osso and Carpita26, Reference Sucksmith, Roth and Hoekstra27 Moreover, in recent years, growing attention has been posed to the evaluation of subthreshold autistic features, such as impaired social and communication skills, unusual and detached personality, and narrow or very intense interests that may represent vulnerability factors for the development of mental disorders.Reference Baron-Cohen, Wheelwright, Skinner, Martin and Clubley28-Reference Takara and Kondo31 Those manifestations have been recognized to be distributed in the general population along a continuum of intensityReference Carpita, Nardi and Bonelli32, Reference Suzuki, Miyaki, Eguchi and Tsutsumi33 and particularly represented in patients with other mental conditions, typically worsening the severity and the course of illness.Reference Dell’Osso, Nardi and Bonelli34, Reference Katsuki, Tateno and Kubo35 Although there are several similarities between ASD and hikikomori, such as deficits in social reciprocity and social withdrawal and sometimes narrow interests, their association has been so far scarcely investigated. Interestingly, some studies have highlighted that subjects with hikikomori have a higher prevalence of ATs such as poor social skills, deficits in communication and imagination, lower attention spans, and lower multitasking and adaptation abilities.Reference Shimono, Hasegawa, Tsuchihara, Tanaka, Matsuda and Kunisato36 Moreover, the dysfunctional social interaction typical of ASD can lead to a reduction in self-esteem and consequently predispose to social withdrawal.Reference Shimono, Hasegawa, Tsuchihara, Tanaka, Matsuda and Kunisato36, Reference Limone, Ragni and Toto37
Another condition that is frequently present in hikikomori subjects is video game addiction. Although video game addiction still lacks a univocal definition and standardized criteria, in the past few decades, it has gained increasing interest as an emerging behavioral dependance, mostly correlated to the rising availability of video games via the internet.Reference Fauth-Bühler and Mann38, Reference Rosendo-Rios, Trott and Shukla39 Indeed, with the enhancement of internet’s expansion and development, it has been reported a noticeable and steady rise in the quantity of video games available online and, therefore, of video game players. Taking into account the most recent statistics, this growth rate jumped from 1.8 billion people regularly playing video games on computers in 2014 to more than two billion in just six years.Reference Krossbakken, Pallesen and Mentzoni40, Reference Vuong, Ho and Nguyen41 In particular, video game addiction combines well with hikikomori syndrome as both share some primary motivations and characteristics drive to use video games in order to escape from real-life problems and to fulfil the need of being good at something.Reference Rosendo-Rios, Trott and Shukla39, Reference Dell’Osso, Amatori and Muti42 Normally these needs can be met through social interaction which hikikomori subjects tend to flee; in this context, video games can represent an effective substitute for in-person social experiences.Reference Rosendo-Rios, Trott and Shukla39 However, while compensating for communication deficits, the use of these means to satisfy the need to interface with others may favor the development of a progressive spiral which ultimately leads to the worsening of both social withdrawal and gaming addiction itself.Reference Yuen, Yan, Wong, Tam, So and Chien43
Interestingly, recently also emerged a link between hikikomori and altered eating behaviors. Indeed, in these subjects, it has been documented that a progressive dysregulation of the eating pattern with an increase in the risk of developing obesity on one side, or extremely restrictive behaviors and underweight on the other.Reference Carpita, Marazziti, Palego, Giannaccini, Betti and Dell’Osso44 This data is of particular interests especially considering the increasing awareness about the association between eating disorders and autism spectrum.Reference Carpita, Muti, Cremone, Fagiolini and Dell’Osso45, 46
To date, studies investigating the association of autistic features and hikikomori are very limited and the studies investigating the presence of a significant overlap between hikikomori and videogame addiction are even fewer. Moreover, although the presence of maladaptive eating behaviors or full-blown eating disorders has been highlighted in hikikomori subjects, no study has systematically evaluated this association with psychometric instruments tailored to assess eating disorder symptoms. In this framework, the aim of this study was to evaluate the presence and correlates of hikikomori tendencies in an Italian University population, focusing on the relationships with autism spectrum, eating disorders, and pathological computer gaming. The choice of this population was motivated by the high rate of dropouts from University students registered in Italy, ranging from around 14% in the first two years up to around 24% after six years. Moreover, only around 25% of students reach their degree in the scheduled time.Reference Uchida and Norasakkunkit47 This elevated drop-out rate might be partially due to the transition from the high school environment, where teaching activities are rigidly organized and attendance of in-person lessons is mandatory, to a university environment where students are autonomous in attending courses and managing oral exams. Moreover, difficulties in finding employment, even after University studies, due to economic stagnation may lead to an increased feeling of hopelessness and fear of unfulfillment, which may undermine the determination in pursuing the chosen academic path. Noticeably, an increase in the so-called “Not in Employment Education or Training” (NEET) phenomenon, which share with hikikomori the tendency to social isolation, unemployment, and lack of sense of belonging was reported in several high-income Countries, and particularly in Italy, with a socio-economic marginalization due to lack of employment also among highly educated people.Reference Orsolini, Longo, Bellagamba, Kato and Volpe48-Reference Bowker, Bowker, Santo, Ojo, Etkin and Raja50 In a previous study, we described two cases of Italian young men who interrupted their University studies due to the development of a full-blown hikikomori condition.Reference Carpita, Bonelli and Giovannoni51 In this framework, considering that some authors have suggested that a socioeconomic environment of insecurity and discouragement regarding future prospects may be a factor facilitating the development of hikikomori,Reference Neoh, Carollo, Lim and Esposito52 it is therefore possible that Italian University would show a greater vulnerability to social withdrawal.
Materials and methods
Participants
An email was sent to all the students of University of Pisa inviting to participate in the study. After consenting to participate, subjects were asked to complete a form that collected sociodemographic data and answer to self-report psychometric instruments. All procedures were conducted through an anonymous online form. Participants did not receive money compensation or other benefit in order to participate in the study. All students were given the possibility to request an interview with a psychiatrist to delve deeper into the results obtained from the various questionnaires.
Psychometric instruments
Hikikomori questionnaire – 25 (HQ – 25)
The HQ-25 is a self-report psychometric instrument that evaluates the severity of hikikomori symptoms during the previous six months. The questionnaire is divided into three domains: socialization, emotional support, and isolation. Answers are organized in a 5-point Likert scale ranging from 0 to 4. In the validation study, a threshold value of 42 was identified for distinguishing between people who were at risk for hikikomori and those who were not, with good sensitivity and specificity.Reference Teo, Chen and Kubo53 An Italian version has been recently validated.Reference Amendola, Presaghi, Teo and Cerutti54
Adult autism spectrum questionnaire (AdAS spectrum)
The AdAS Spectrum is a self-report questionnaire developed to determine the presence of full- and subthreshold autistic symptoms and traits in adults without intellectual disability.Reference Dell’Osso, Gesi and Massimetti55 The 160 items in the questionnaire are categorized into 7 domains: Childhood/adolescence, Verbal communication, Nonverbal communication, Empathy, Inflexibility and adherence to routine, Restrictive interests, and rumination and Hyper-hypo reactivity to sensory input. The tool has two validated threshold values: a cutoff score of 43 to indicate the presence of clinically relevant ATs and a cutoff score of 70 for full-blown ASD symptoms.Reference Dell’Osso, Carmassi and Cremone56
Eating attitude test – 26 (EAT – 26)
The EAT-26 is a self-report instrument used for the screening of various manifestations associated with eating disorders, developed for both clinical and nonclinical samples. The questionnaire is divided into three domains exploring dieting, bulimia and food preoccupation, and oral control. During its validation study, the instrument reported excellent psychometric properties and great reliability and validity.Reference Garner, Olmsted, Bohr and Garfinkel57 A cutoff of 20 was considered the threshold for the possible presence of an eating disorder.
Assessment of internet and computer game addiction (AICA – S)
The AICA-S is a self-report questionnaire that investigates the possible presence of videogaming addiction. The instrument includes 14 items organized in 5-point Likert scale. Scores of 7 or above are indicative of pathological computer game use, while scores between 4 and 7 suggest the presence of an excessive computer game use.Reference Curcio, Peracchia and Presaghi58
Statistical analysis
Statistical evaluations were performed with SPSS version 26.0. The sample was divided into two groups, one at risk for Hikikomori and one not at risk, considered as healthy controls (HCs) based on the HQ-25 cutoff score. A chi-square analysis was used to compare the presence of a risk for hikikomori in the overall sample, based on gender. A student’s t-test was used to compare AdAS spectrum, EAT-26, and AICA responses between the groups. Three chi-square analyses were performed to compare the presence of significant ATs, full-blown ASD, moderately and pathological addictive gaming, and eating disorder symptoms in the two groups. Pearson’s correlation coefficient was then used to evaluate the pattern of correlations between the scores obtained in all questionnaires’ domains and totals in the overall sample. A factorial analysis of variance (ANOVA) was carried out to evaluate how the presence of an eating disorder, full-blown ASD and pathologically addictive computer game use interacted to affect the HQ-25 score. A logistic regression analysis was performed using the presence of hikikomori risk according to the HQ-25 cut-off as a dependent variable and AdAS spectrum, EAT-26, and AICA-S total scores as independent variables. Similarly, we performed a linear regression analysis using the same independent variables, with HQ-25 score as a dependent variable, to further confirm the previous results. Further logistic regressions were then carried out using the presence of hikikomori risk as a dependent variable and AdAS spectrum and EAT-26 domains as independent variables. Subsequently, a decision tree model was performed with the aim to identify which variables among the presence of significant AT/full-blown ASD and the presence of excessive/pathological computer game use best predicted belonging to the group at risk for hikikomori. The chi-squared automated interaction detection (CHAID) growing method was used. With the use of this analysis, we were able to evaluate how different factors interact and build a decision tree model that visually represents the results as an inverted tree. The model begins with a root node that contains all the cases. Next, the tree is built by identifying the key discriminating variables: at each step, CHAID chooses the independent variable that shows the strongest interaction with the dependent variables. When there is no reported difference with respect to the dependent variable, the model also combines the categories described by the predictors.
Results
The total sample was made of 2574 university students: 1175 (45.6%) males, 1345 (52.3%) females, and 54 (2.1%) nonbinary subjects; the mean age of the overall sample was 25.02 years (±6.28).
Based on the threshold value of the HQ, 1047 (40.7%) subjects fell into the group at risk for Hikikomori while 1527 (59.3%) did not report a pathological score.
A chi-square analysis showed significant differences in the presence of risk for hikikomori based on gender, highlighting a higher proportion of subjects at risk of hikikomori among nonbinary subjects, an intermediate proportion among males and the lowest proportion among females (see Table 1).
Table 1. Comparison of frequency of the risk for hikikomori in the overall sample based on gender
Each subscript letter denotes a subset of Diagnostic categories whose row proportions do not differ significantly from each other at the,05 level.
The t-student results reported in Tables 2, 3 and 4 highlighted how the group at risk for hikikomori showed significantly higher scores in all AdAS spectrum, AICA-S, EAT-26 domains, and total score. Figure 1 represents a profile plot of estimated z-scores reported on AdAS spectrum, AICA-S, and EAT-26 totals depending on the presence/absence of Hikikomori risk.
Table 2. Comparison of AdAS Spectrum scores between the group at risk for hikikomori and HCs
Table 3. Comparison of EAT-26 scores between the group at risk for hikikomori and HCs
Table 4. Comparison of AICA total scores between the group at risk for hikikomori and HCs
Figure 1. Profile plot of estimated z-scores reported on AdAS spectrum, AICA-S, and EAT-26 totals depending on the presence/absence of Hikikomori risk.
Results from the first chi-square analysis showed that subjects with significant ATs and with clinical ASD symptoms were statistically more represented in the group at risk for hikikomori compared to the HCs, while, in turn, subjects with clinical symptoms of ASD were more represented in the hikikomori group than subjects with significant ATs (see Table 5).
Table 5. Comparison of the presence of significant ATs or putative ASD in the group at risk for hikikomori and the HCs
Each subscript letter denotes a subset of Diagnostic categories whose column proportions do not differ significantly from each other at the ,05 level.
The second chi-square analysis, described in Table 6, showed that subjects with pathological or excessive computer game use were statistically more represented in the group at risk for hikikomori than HCs, while subjects with excessive computer game use were more represented in the hikikomori group also compared with subjects with excessive video game use.
Table 6. Comparison of the presence of addictive internet use in the group at risk for hikikomori and the HCs
Each subscript letter denotes a subset of Diagnostic categories whose column proportions do not differ significantly from each other at the ,05 level.
The third chi-square analysis (see Table 7) highlighted that subjects with eating disorder symptoms were statistically more represented in the group at risk for hikikomori.
Table 7. Comparison of the presence of eating disorder symptoms according to EAT-26 in group at risk for hikikomori and the HCs
Each subscript letter denotes a subset of Diagnostic categories whose column proportions do not differ significantly from each other at the ,05 level.
The correlation analysis showed that, in the total sample, AdAS spectrum, HQ-25, AICA-S, and EAT-26 total and domain scores were all positively and significantly correlated. However, while the correlations reported between HQ-25 and AdAS spectrum total and domain scores were moderate to strong, those between HQ-25 and AICA-S were moderate to weak and those between AICA-S and EAT-26 were very weak (Tables 8 and 9).
Table 8. Pearson’s correlations coefficients (r) between AdAS spectrum domains and total scores and HQ and EAT-26 domains and total scores and AICA-S total score
* p < .05
Table 9. Pearson’s correlations coefficients (r) between HQ domain and total scores, EAT-26 domain, and total scores and AICAS total
* p < .05
The ANOVA analysis showed a significant main effect of the presence of eating disorder symptoms, ATs/possible clinical ASD, and pathological/excessive videogame use on HQ-25 total scores (see Table 10). However, while all variables produced a significant separate effect on HQ-25 scores, no significant interaction was found between each other in this effect.
Table 10. Factorial ANOVA analysis with the presence of eating disorder symptoms, AT/full-blown ASD and excessive/pathological computer game use as independent variables and HQ-25 total score as dependent variable
R squared = 0.260; Adjusted R squared = 0.255. Significant p values are reported in bold.
The logistic regression, shown in Table 11, highlighted AdAS spectrum and AICA-S total scores as significant predictors of belonging to the group at risk for hikikomori. Those results were confirmed by a subsequent linear regression analysis (see Table 12).
Table 11. Logistic regression analysis with the HQ-25 above-threshold scores as a dependent variable and AdAS spectrum, EAT-26, and AICA-S total scores as independent variables
Cox & Snell R square = 0.187; Nagelkerke R square = 0.302
Table 12. Linear regression analysis with HQ-25 above-threshold scores as a dependent variable and AdAS spectrum, EAT-26, and AICA-S total scores as independent variables
R square = 0.303; Adjusted R square = 0.421
Lastly, two logistic regression analyses reported EAT-26 Oral control domain (see Table 13) and all AdAS spectrum’s domains with the exception of hyper hyporeactivity to sensory input and inflexibility and adherence to routine (see Table 14) as significant positive predictors of being at risk for hikikomori. AdAS spectrum domain Inflexibility and adherence to routine was instead reported to be a negative predictor of the risk for Hikikomori.
Table 13. Logistic regression analysis with the presence of hikikomori risk according to HQ-25 cutoff as a dependent variable and EAT-26 domains as independent variables
Cox & Snell R square = 0.032; Nagelkerke R square = 0.043.
Table 14. Logistic regression analysis with the presence of hikikomori risk according to HQ-25 cutoff as the dependent variable and AdAS spectrum domains as independent variables
Cox & Snell R square = 0.217; Nagelkerke R square = 0.293
The decision tree model, performed using HQ-25 cutoff as a dependent variable and the presence of significant ATs or ASD clinical symptoms, the presence of excessive or pathological videogame use and the presence of eating disorder symptoms as independent variables, showed in the first step a significantly higher risk for hikikomori among both subjects with significant ATs and ASD clinical symptoms. Subsequently, among both subjects with significant ATs or ASD clinical symptoms, the possibility of having a relevant risk for hikikomori was higher in subjects with pathological videogame use compared to those with excessive or no addictive videogame use. No difference was reported depending on the presence of eating disorder symptoms (see Figure 2).
Figure 2. Decision tree model with HQ-25 cutoff as a dependent variable and the presence of significant ATs or ASD clinical symptoms, the presence of excessive or pathological videogame use and the presence of eating disorder symptoms as independent variables
Discussion
In this study, we aimed to assess the prevalence of a hikikomori symptomatology in University students, as well as its correlation and relationship with some major psychopathological dimensions such as autistic features, altered eating behaviors, and pathological videogame use.
Based on the HQ-25 threshold score, the prevalence of hikikomori risk in the sample recruited appeared to be strikingly high compared to the rates usually described in nonclinical populations.Reference Eckardt59 However, since the recruitment was carried out on a voluntary basis, it is possible to assume that people who identified with the characteristics described were motivated to join and complete the tests. On the other hand, as previously stated, the Italian University student population may be at higher risk of hikikomori tendencies with respect to the general population. School abandonment during University is quite high in Italy, as well as the rate of NEET in youths, while we recently reported a case series of young men who interrupted their University studies due to the development of a full-blown hikikomori condition.Reference Bowker, Bowker, Santo, Ojo, Etkin and Raja50, Reference Quintano, Mazzocchi and Rocca60 In the context of socioeconomic crisis, the lack of expectations for future perspectives is dramatically spreading among young people. This issue, together with the challenges of an academic path which requires, compared to Italian high school, high levels of effort and autonomous organization, without assuring work possibility after the degree, may lead to a sense of insecurity and discouragement regarding the future, which is considered a risk factor for the development of hikikomori behaviors in vulnerable subjects.Reference Orsolini, Bellagamba, Volpe and Kato49, Reference Bowker, Bowker, Santo, Ojo, Etkin and Raja50, Reference Neoh, Carollo, Lim and Esposito52
Our results about gender differences are consistent with other studies that described a greater presence of hikikomori in males than femalesReference Nonaka, Takeda and Sakai61; however, we found an even higher prevalence in nonbinary individuals. To our knowledge, this is the first study to evaluate the presence of hikikomori in nonbinary subjects; the results obtained are in line with the presence of greater psychopathology, including the ATs, reported in this population in previous studies.Reference Kung62-Reference Newcomb, Hill, Buehler, Ryan, Whitton and Mustanski64
Our results highlighted how subjects at risk for hikikomori were more likely in the autism spectrum than HCs, showing also more frequently maladaptive computer game habits and eating disorder symptoms. This evidence is in line with the available literature that highlighted a greater prevalence of ASD features in subjects suffering from hikikomori,Reference Shimono, Hasegawa, Tsuchihara, Tanaka, Matsuda and Kunisato36, Reference Yuen, Yan, Wong, Tam, So and Chien43, Reference Brosnan and Gavin65, Reference Yamada, Kato and Katsuki66 as well as higher scores in questionnaires investigating internet/gaming addictionReference Dell’Osso, Amatori and Muti42, Reference Stavropoulos, Anderson, Beard, Latifi, Kuss and Griffiths67, Reference Kubo, Horie and Matsushima68 and a greater prevalence of altered eating habits in said population.Reference Yuen, Yan, Wong, Tam, So and Chien43 As shown in Figure 1, the greatest differences between subjects with or without hikikomori tendencies were found on the AdAS Spectrum score, secondly on the AICA-S score and finally on eating disorders. To our knowledge, although a possible association between maladaptive eating behaviors is reported in the literature, such as irregularity in eating or the tendency towards neglect and wasting,Reference Kato, Kanba and Teo14 this is the first study which systematically investigates the association between hikikomori and the eating disorder spectrum through validated questionnaires. The association between hikikomori and altered eating behaviors could be explained by the core hikikomori features, such as the refusal to leave one’s room and the often frequent inversion of circadian rhythms, which can lead to irregularities in the subject’s eating pattern. Furthermore, hikikomori subjects, due to their severe social withdrawal, avoid leaving their homes even to carry out basic necessities such as buying food. For this reason, it is likely that these subjects may go for several days without adequate nourishment, that they may eat long-perishable goods that do not satisfy the characteristics of a balanced meal or that they abuse fast-food services with home delivery with related consequences on your body.Reference Yuen, Yan, Wong, Tam, So and Chien43 Interestingly, a recent study has described how up to 70% of subjects with hikikomori were either underweight and overweight/obese, with underweight being more frequent among new-onset cases.Reference Yuen, Yan, Wong, Tam, So and Chien43 Our findings highlighted significant and positive correlations between all the psychometric instruments employed, with the highest reported among HQ-25 and AdAS Spectrum, strengthening our previous results. However, moderate correlations were also reported with AICA-S, for both HQ-25 and AdAS spectrum. The correlation between autistic traits and the pathological use of video games, which generally involve strong online involvement, can be explained as a manifestation of a specific narrow interest in computer games, but also by the reduced ability in social interaction and difficulties in building interpersonal relationships typical of ASD subjects. The internet could be considered an idealistically safe communicating place for subjects with autistic traits, especially when communicating indirectly through gaming, to the point that it can become the only method of interacting with reality.Reference Murray, Koronczai, Király and Autism69 Similarly, autistic-like social deficits may enhance a pathological social withdrawal, being a vulnerability factor towards the development of hikikomori behaviors.Reference Dell’Osso, Amatori and Muti42 In this last case, some researchers hypothesized a possible common neurobiological basis between ASD and hikikomori highlighting the importance of neurodevelopmental alterations. This theory is supported by the innate tendency that subjects with high autistic traits have towards isolation, called by some authors ‘hikikomori affinity’, and that sometimes can anticipate the full-blown manifestation of a hikikomori syndrome.Reference Dell’Osso, Amatori and Muti42, Reference Brosnan and Gavin65, Reference Kato, Shinfuku and Tateno70
Considering EAT-26, the highest correlations were found with AdAS spectrum, while, although significant, the correlations highlighted with HQ-25 and AICA-S, were quite weak, identifying them as the dimensions least associated with each other among those examined. The link between autistic traits and eating disorders is widely recognized and described in the current literature, especially among females.Reference Baraskewich, von Ranson, McCrimmon and McMorris71-Reference Suarez73 The hypothesis of a possible correlation between the two disorders is fueled by the evidence of a greater presence of some typically autistic traits in patients with Feeding and Eating Disorders (FEDs) and in parallel altered eating behaviors in ASD subjects.Reference Carpita, Muti, Cremone, Fagiolini and Dell’Osso45, Reference Barrionuevo, Chowdhury and Lee74-Reference Westwood, Eisler, Mandy, Leppanen, Treasure and Tchanturia76 On the other hand, the presence of a link between eating disorders and pathological use of video games as well as extreme social withdrawal is in line with some emerging evidences. While studies focusing on the specific relationship between eating disorders and videogame use are still scant, some authors are highlighting poorer dietary habits associated with video gaming.Reference Lebby, Shyam and Ramadas77, Reference Puolitaival, Sieppi, Pyky, Enwald, Korpelainen and Nurkkala78 In particular, a population-based study described how adolescent who engage extensive time in playing videogames were more likely to have poorer dietary habits compared to their peers.Reference Puolitaival, Sieppi, Pyky, Enwald, Korpelainen and Nurkkala78 Those results were confirmed by a later study that underlined how the frequency of videogame use was associated with significantly higher BMI as well as obesity and diabetes.Reference Lebby, Shyam and Ramadas77 This phenomenon could be explained by the fact that the pathological use videogames, which involves sitting for much of the day, would favor the appearance of a sedentary lifestyle and therefore obesity and binge-eating behaviors. On the other hand, modern video games often involve matches that last several hours and that could ultimately lead to neglecting or skipping regular meals.
The factorial ANOVA analysis highlighted a significant main effect of the presence of an eating disorder, ASD, and videogame use on the HQ-25 score, but failed to find any significant interaction between them, highlighting how their effects on hikikomori risk are independent from one another. Among these variables, the strongest significant factor appeared to be the presence of an autism spectrum, followed by the maladaptive use of videogames and lastly the presence of an eating disorder. These results highlight the importance of the autistic dimension as the main discriminant for hikikomori risk. Such a link between the autistic spectrum and hikikomori was further confirmed by the results of the regression analyses that underlined how AdAS spectrum total score and the majority of its domains (in particular, all the domains investigating social and communication difficulties, such as empathy and verbal/nonverbal communication issues) emerged as significant predictors of a hikikomori symptomatology. This result supports the hypothesis that the presence of significant autistic traits, or of full-blown ASD, could be a vulnerability factor for the development of hikikomoriReference Shimono, Hasegawa, Tsuchihara, Tanaka, Matsuda and Kunisato36, Reference Dell’Osso, Amatori and Muti42 and that the core difficulties in interpersonal relationships and social communication of subjects in the autism spectrum may lead them to social isolation, up to its extreme pathological drift, the hikikomori syndrome.Reference Shimono, Hasegawa, Tsuchihara, Tanaka, Matsuda and Kunisato36, Reference Dell’Osso, Amatori and Muti42, Reference Brosnan and Gavin65, Reference Kato, Shinfuku and Tateno70 In fact, although studies in the field are still scant, some authors have highlighted how the presence of autistic features, in particular the difficulties in communication and social interaction were predictors of hikikomori risk.Reference Brosnan and Gavin65, Reference Kato, Katsuki and Kubo79-Reference Mazurek, Shattuck, Wagner and Cooper81 According to our data, another significant hikikomori predictor was the presence of autistic manifestations in childhood/adolescence, such as the difficulty in fitting into a group of friends or making new ones, the predilection for solitary activities and being bullied, all features that represent a first alarm towards the tendency to social isolation.Reference Wakuta, Nishimura and Osuka82, Reference Ding and Zhang83 Previous studies described how subjects with hikikomori are usually individuals refusing social contact with others since childhood and episodes of childhood bullying.Reference Krieg and Dickie84 Finally, our findings highlighted also the autism spectrum dimension of ruminative thinking and restricted interests as a positive predictor of hikikomori risk: the propensity towards the intense persecution of narrow interests, which in the case of hikikomori mainly revolve around online and solitary activities, allows the maintenance and worsening of the isolation framework.Reference Brosnan and Gavin65 Moreover, negative ruminative thinking, increasing the focus on feelings about problems rather than on problem-solving, may enhance the development of maladaptive strategies for copying with social difficulties, including social withdrawal.Reference Dell’Osso, Cremone and Amatori85 Interestingly, we also observed a protective role of the Inflexibility and adherence to routine domain towards hikikomori symptomatology. This data could be explained by assuming that rigidity, as an autistic trait, could prevent the subject from locking himself up at home in order to maintain his routine in his daily habits. This data may be also in line with the protective role of this dimension with respect to suicidality and for the development of Borderline Personality Disorder (BPD) symptoms, possibly due to a counterbalance of impulsivity-related issues.Reference Dell’Osso, Cremone and Amatori85, Reference Carpita, Bonelli and Schifanella86
According to our results, the AICA-S total score also emerged as a significant predictor for a higher hikikomori symptomatology, which also aligns with the aforementioned evidences. Indeed, while social deficits linked to hikikomori can lead to a decrease in interpersonal relationships and social withdrawal, on the other hand, the use of videogames can become the only means of interaction with the outside world, a compensatory tool, which would however enhance its pathological use, worsening, in a vicious cycle, the social withdrawal.Reference Bowman, Rieger and Tammy Lin87-Reference Stip, Thibault, Beauchamp-Chatel and Kisely91 Moreover, some studies seemed also to highlight that hikikomori subjects have greater satisfaction from virtual relationships compared to real ones.Reference Stip, Thibault, Beauchamp-Chatel and Kisely91
Interestingly, EAT-26 total score did not emerge as a significant predictor of hikikomori but only the EAT-26 Oral Control domain. The Oral Control domain analyzes the degree of self-control over eating and the perception of direct external pressures on modifying one’s body weight by increasing it. Individuals with higher scores in this domain may fear the external pressures aimed at favoring weight gain or regulating their dietary habit, perceiving the external environment as potentially dangerous towards their eating behavior. As nowadays food involves many aspects of social events, this particular feature can ultimately lead to an avoidance of partaking social activities and to a preference for social isolation that would allow the subject to eat as they like without having to suffer external pressure.Reference Herman92 The above-mentioned results have been further confirmed by a decision tree model analysis that showed how the presence of subthreshold or full-threshold ASD symptoms was the most discriminant dimension with respect to the presence/absence of hikikomori. Secondly, within the subgroups with ATs or with possible symptoms of ASD, the AICA-S pathological threshold was discriminant for belonging to the hikikomori group. From a dimensional perspective in psychopathology, in which the concept of spectrum is introducedReference Baron-Cohen, Wheelwright, Skinner, Martin and Clubley28 and ATs are considered to be distributed along a continuum in the general and clinical population, the social withdrawal of hikikomori individuals could be hypothesized to be underlain by autistic features.Reference Shimono, Hasegawa, Tsuchihara, Tanaka, Matsuda and Kunisato36, Reference Brosnan and Gavin65, Reference Yamada, Kato and Katsuki66
These results should be considered in light of important limitations. The sample recruited included only University students preventing the generalization of the results to the general population. Secondly, subjects were chosen on a voluntary basis, and possible biases related to sample selection should be taken into account, such as an over-representation of subjects more interested in the issue. Thirdly, we used self-reported questionnaires, which may imply the risk of over−/under-estimation of symptoms. Lastly, due to the cross-sectional design of the study, we are unable to draw conclusions about casual or temporal links among the investigated variables.
Globally, our findings support the association among hikikomori, autism spectrum, pathological computer game use, and eating disorder symptoms. From a broader viewpoint, our findings may be in line with the rising hypothesis of a possible neurodevelopmental basis for different psychiatric conditions, including, as in the case of this study, hikikomori.Reference Dell’Osso, Lorenzi and Carpita93
Data availability statement
All data generated or analyzed during this study are included in this published article.
Acknowledgments
Not applicable.
Author contribution
Conceptualization: B.C., I.M.C., S.P., L.D.O; Methodology: B.C., I.M.C., S.P., E.P., L.D.O.; Formal analysis: B.C., G.M.; Investigation: B.C., B.N., C.B., G.A.; Supervision: B.C., I.M.C., S.P., L.D.O.; Writing original draft: B.N., F.G., F.P., G.C.; Writing review and editing: B.N., B.C.
Funding statement
This research received no external funding.
Competing interest
The authors declare no conflict of interest.
Ethics approval statement
All procedures were approved by Committee on Bioethics of the University of Pisa Review No. 18/2023 on May 26th, 2023.
Patient consent statement
All recruited subjects consented to participate in the study.
Open access
