This was the first study that examined a longitudinal relationship between DIS and SH in the general adolescent population. While DIS tended to predict future SH, SH did not predict future DIS. Individuals with persistent SDIS had an approximately three times higher risk of SH than those with no experience of SDIS. Individuals with persistent SH tended to have a higher risk of SDIS.

The results of this study showed that DIS tended to longitudinally predict future SH. This result is in line with a previous study that showed that a decrease in DIS predicted a decrease in SH in adolescent female victims of sexual abuse [26]. Furthermore, regarding the temporal relationship, the results also agree with previous cross-sectional studies that showed that DIS mediated the relationship between childhood trauma and SH [8, 11, 14, 15, 24] Regarding the empirical and theoretical hypotheses, this longitudinal relationship may support the anti-dissociation model of SH [20]. In previous studies, approximately half of adolescents with SH endorsed the reasons that were assumed to end DIS, such as “to stop feeling numb or out of touch with reality” or “to feel something even if it is pain” [39, 40] A previous study suggested that feeling pain or seeing their blood may be instrumental in retaining reality or a coherent sense of existence and ending DIS. [41] Another explanation is that this longitudinal relationship may also support the self-punishment model of SH. [20] A previous study suggested that SH may be triggered by intrapsychic conflict in severely dissociated persons, in which one dissociated part could be aggressive to another. [21] In both hypotheses, temporal relief after SH as anti-dissociation or self-punishment may reinforce SH and lead to its repetition. [42]

We also revealed that those with persistent SDIS had a significantly increased risk of SH at T2. This suggests that the repetition of SDIS over a 2-year period may lead to the initiation or maintenance of SH. To the best of our knowledge, no study has examined the association between persistent SDIS and SH in the general population of adolescents. While a previous study suggested that the severity and frequency of DIS are associated with the severity of SH, [25, 26] attention should also be given to the persistence of SDIS as a risk factor for SH. Since the transient SDIS group did not have an increased risk for SH, spontaneous transient SDIS may not increase the risk for SH. This result may also suggest the importance of early intervention for preventing future SH in adolescents with SDIS [43].

This study revealed that SH did not predict DIS 2 years later. Although a previous study suggested that SH is a deliberate attempt to dissociate and escape from unbearable distress, [22] SH may not promote a long-lasting tendency to dissociate. Nonetheless, it should be noted that adolescents with persistent SH tended to have an increased risk of SDIS. This is in line with previous findings that habitual SH was associated with DIS [14, 44] although the associations were stronger in previous studies, probably due to the cross-sectional design. This study also revealed that adolescents with incident SH had a significantly higher risk of SDIS. Therapists treating adolescents with SH should pay attention to the comorbidity of SDIS, especially if the SH is persistent or relatively recently initiated.

Although the cross-sectional association between DIS and SH was significant, the longitudinal relationship between DIS and SH was weak or non-significant. The observed significant cross-sectional association between DIS and SH fits with previous findings [6,7,8,9,10,11,12,13,14,15, 26]. There may be several explanations for the relationship between DIS and SH that was observed in this study. First, DIS and SH may have a longitudinal relationship over a relatively short period. Since we assessed DIS and SH twice over a relatively long period (2 years), the longitudinal relationship between them might be observed to be weaker. Second, DIS and SH may mostly co-occur rather than one leading to the other. DIS and SH may both occur in adolescents who suffer from acute strong distress. In addition, they may also occur in adolescents who suffer from strong distress when recalling past adverse experiences. For example, childhood trauma is a risk factor for both DIS and SH [11, 14, 15, 36], and a previous study showed that the majority of dissociative disorder patients had self-harmed after trauma-related cues [45].

Several clinical implications can be drawn from this study. First, interventions for DIS should be considered to prevent future SH in adolescents, even if they do not currently engage in SH. The interventions may focus on distress, which promotes DIS. [46] Second, intensive attention should be given to adolescents who have repeated SDIS since they are at a specifically higher risk of SH. Persistent SDIS may be regarded as a representation of unresolved persistent distress.

The strengths of this study include the design of the TTC study. First, this study was longitudinal with a prospective design. Since most previous studies were cross-sectional, the results of our study may contribute to the understanding of the theoretical relationship between DIS and SH. Second, the sample size of this study was relatively large compared to previous cross-sectional studies with general populations [6,7,8,9,10]. Third, this study used a sample of adolescents from the general population. Since only a small portion of adolescents receive medical treatment for SH, [1] this study may contribute to the development of a generalizable prevention strategy.

However, there are some possible limitations. First, we assessed DIS by the parent-report CBCL. There may be concerns about the validity of a parent’s assessment of a child’s internal experience. The parent’s assessment might underestimate or overestimate DIS. Only DIS with more than a substantial severity might be captured by parents, but mild DIS might be missed and underreported. On the other hand, contemplation or meditation might be regarded as DIS by parents and could be over-reported. Since most previous cross-sectional studies on the relationship between DIS and SH used self-report assessment of DIS [6,7,8,9,10,11], our results should be compared with previous studies carefully. Second, since we assessed SH only by children’s self-reports, the possibility of underreporting should be considered; however, a previous study suggested that self-reports were suitable for highly sensitive questions [47]. In addition, we used one generic question to assess SH at T1, although we assessed SH at T2 in more detail using free description. Although the measurement method was similar to that of previous studies[2, 3, 48, 49], it is a limitation that we did not use validated questionnaires to assess SH. Third, most of the participants were Japanese adolescents living in Tokyo, an Asian metropolis. Careful consideration is needed to generalize these study findings to other ethnic groups and countries. Fourth, we did not obtain information about childhood trauma and abuse from family members because asking about such experience was considered possibly invasive for the adolescents. This may limit the interpretation of the results of this study in conjunction with other previous works. Both DIS and SH were suggested to be associated with childhood trauma [11, 14, 36], while a previous study revealed that a higher level of DIS was related to SH even after controlling for childhood abuse [50]. Further study is needed to examine the longitudinal relationship among DIS, SH, and childhood trauma. Future studies are also warranted to investigate the longitudinal relationship between DIS and SH over a shorter period.