Our study compared ITS and DSH participants with controls to investigate the differences between ITS and DSH. Our SEMs showed that parental bonding had the greatest influence on the development of DSH behavior, with paternal bonding having a greater influence than maternal bonding; neither DIF nor mental health had a direct influence on DSH. On the other hand, parental bonding did not have a direct influence on ITS; instead, an indirect effect of maternal bonding on ITS was found through the mediating factor of the alexithymic trait of DIF. Participants who were younger, less extraverted, and with a higher DIF alexithymic trait and worse mental health were more likely to develop ITS behavior. In these models, given that DSH accounted for 65% of the variance and ITS accounted for 52% of the variance, it can be concluded that DSH and ITS differ psychopathologically.
No demographic differences were found between the DSH and the control group; on the other hand, two demographic factors, age and gender, were directly or indirectly associated with ITS. Our results showed that age was directly associated with ITS; those who were older were less likely to show ITS behavior. In Taiwan, suicide has been reported to be the second most common cause of death among adolescents (15–24 year olds) . This shows that adolescents have the highest risk of ITS in Taiwan, although they are younger than subjects included in our sample. In addition to age, gender had an indirect association with ITS through the mediating factor of the DIF factor of alexithymia. Males were more likely to be alexithymic (as shown in Figure 2). This is consistent with a previous study, which found males had higher scores for alexithymic traits than females . Therefore, those who are male and have difficulty identifying their feelings, and those who are younger, are at higher risk of developing ITS behaviors than other members of the population.
Beside demographics, parental bonding was found to be associated with both ITS and DSH. The bonding to both parents was directly associated with DSH, and maternal bonding was indirectly associated with ITS through the mediating factor of the alexithymic trait of DIF. An influence of parental bonding on suicidal behavior was also found in previous studies, which showed that parental bonding was associated with the repetition of suicidal behavior , and with suicidal ideation and attempts in adolescents . Alexithymics have recalled a parenting style of either overprotection or low care, which showed an association between parental bonding and alexithymic traits [46, 47]. A similar pathway, with personality characteristics being the mediating factor between parental bonding and alexithymic traits, has also been found . Furthermore, our study revealed, further, that paternal bonding had a greater association with DSH than maternal bonding. However, only maternal bonding had an indirect association with ITS; paternal bonding was not associated with ITS. Therefore, although parental bonding influences the development of both ITS and DSH behaviors, the role of paternal bonding is of primary importance in the development of DSH behavior, whereas the role of maternal bonding is of primary importance in the development of ITS behavior.
In addition to the level of DIF, ITS patients also had worse mental health. However, mental health did not influence the development of DSH behaviors. Depression has been associated with suicidal ideation ; feeling hopeless was determined to be a predictor of both suicidal ideation and depression . In adolescents, a high association between depression and anxiety was associated with self-harming behavior in a follow-up cohort study . However, this association was not found in our study in participants aged between 29 and 50 years old. Furthermore, those with intention to commit suicide were not excluded from the study of Moran and colleagues . Given that ITS patients have statistically significantly worse mental health conditions than controls, public and medical health professionals should pay careful attention to patients who have attempted suicide, and refer them for psychiatric help when needed.
The first limitation of this study is related to the fact that a previous study showed that the vast majority of adolescents with DSH behavior do not seek medical help . Given that the participants in this study were recruited from the emergency room of a hospital, our results can only be generalized to those who have sought help at a hospital. Second, the participants in this study were recruited from the emergency room of a general hospital in southern Taiwan. The rejection rate for recruitment was high when there were no medical rapport. In addition, given that we were unable to follow up those who rejected participation, the demographics of those who agreed to participate could not be compared with the demographics of those who declined to participate. Therefore, the external validity of the results needs further investigation. Lastly, Chen et al.  showed that the media tends to underreport mental illness as a reason for suicide in men in Taiwan. In India, the National Crime Records Bureau underestimates suicide in men by at least 25%, and in women by at least 35% . This shows that some gender and cultural factors are associated with the tendency to commit suicide. Accordingly, semi-structured interviews regarding the motivation behind their behavior were conducted with all the patients who agreed to participate to determine whether they had intended suicide or self harm. However, we cannot eliminate the possibility that they underreported their intention to commit suicide. The atmosphere of the emergency room department is generally noisy and chaotic in Taiwan. We suggest in future studies a private room maybe used for these interviews to elevate medical rapport and diminish the affect from the environment. Notwithstanding those limitations, the strength of this study is that it included only first-time DSH and ITS participants without any psychiatric disorder or general medical disorders. Psychiatric inpatients, general medical patients, and community residents who attempt suicide have different psychological risk factors . Therefore, this study excluded psychiatric disorders and other general medical disorders, and focused only on those in the community.